The prognostic value of ultrasound abnormalities and biological parameters in blood of fetuses infected with cytomegalovirus


Prof Y Ville, Service de Gynécologie Obstétrique, Centre Hospitalier Intercommunal de Poissy-St Germain, 10, rue du Champ Gaillard, 78300 Poissy, France. Email


Objective  To evaluate the prognostic value of ultrasound abnormalities and of selected biological parameters in blood of fetuses infected with cytomegalovirus (CMV).

Design  Retrospective observational study.

Setting  Two fetal medicine units in Paris, France.

Population  All fetuses infected with CMV referred between 1998 and 2006.

Methods  We retrospectively analysed data collected prospectively in 73 fetuses infected by CMV with a positive CMV polymerase chain reaction in amniotic fluid. Fetal blood sampling (FBS) was performed for evaluation of platelet count, plasma levels of aminotransferases and gamma-glutamyl transpeptidases (GGT), presence of viraemia and specific fetal immunoglobulin M. Targeted ultrasound examination was performed every fortnight. Ultrasound findings were categorised into normal examination and any ultrasound abnormality, which was further grouped as ultrasound abnormality of the fetal brain and noncerebral ultrasound abnormality.

Main outcome measures  A combination of histological findings after termination of pregnancy and evidence of cytomegalic inclusion disease at birth when pregnancies were continued. Clinical symptoms at birth or histological lesions attributable to CMV were considered as poor outcome. Statistical analysis was conducted to determine the value of each parameter to predict outcome. Logistic regression was used to build up a multivariate model combining the relevant parameters.

Results  In univariate analysis, only thrombocytopenia and the presence of any ultrasound abnormality were associated with a poor outcome (P < 10−4 for both abnormalities). In the multivariate analysis, both thrombocytopenia and the presence of ultrasound abnormalities remained significant independent predictors of a poor outcome. Based on univariate logistic regression, odds ratio for a poor outcome were 1.24, 7.2, 22.5 and 25.5 for each 10 000/mm3 decrease in platelet count, the presence of noncerebral, any ultrasound and cerebral ultrasound abnormalities, respectively.

Conclusions  The prognosis of CMV-infected fetuses relies independently on both targeted ultrasound examination and fetal platelet count. FBS for platelet count may therefore justify FBS in infected fetuses even in the absence of ultrasound. features of brain involvement.


Cytomegalovirus (CMV) is the main cause of viral congenital infection.1–4 In industrialised countries, around 50% of pregnant women are immune for CMV at the beginning of their pregnancy.5–7 Primary infection occurs in 1–2% of the pregnancies with a vertical transmission rate of 30–40%.1,4,8 Recurrent infection may also occur in pregnant women with preconceptional immunity, with a vertical transmission rate of around 1%.9,10 Less than 1% of all newborns are infected with CMV. This is the result of primary maternal infection in 2/3 of the cases and of secondary infection in another 1/3 of the cases.11,12 Primary infection carries an a priori risk of giving birth to a symptomatic infant in 10–15% of the cases.11,13

Symptomatic newborns will show at least one of the following signs: hypotrophy, microcephaly, jaundice, hepatosplenomegaly, thrombocytopenia and related petechiae, seizures, thus defining the cytomegalic inclusion disease (CID).14–17 Around 5% of symptomatic newborns will die and either one of neurological handicap, visual impairment or hearing loss will develop in almost 90% of the infants who survive.18 The prognosis is more difficult to establish in asymptomatic neonates and must take into account the normality of all neonatal investigations. Overall, up to 10–15% of asymptomatic children at birth will develop some degree of (usually unilateral) sensorineural hearing loss and/or learning difficulties during or after the first years of life.11,18,19 The definition of an asymptomatic fetus is likely to include normal fetal growth and brain ultrasound examination, and it could also include normal plasma levels of liver enzymes and platelet count.

Brain abnormalities as well as noncerebral abnormalities are objective ultrasound anomalies more often associated with a poor prognosis. At birth, the cytomegalic inclusion disease is defined by a combination of clinical, imaging and biological abnormalities. In the fetus, hepatitis and thrombocytopenia can only be diagnosed by fetal blood sampling (FBS), but their significance is more hypothetical. Furthermore, there is no correlation between quantification of the CMV genome in the amniotic fluid and neonatal outcome.20 The prognostic value of fetal viraemia or viral load has been evoked but remains controversial.21–23 The aim of this study was to evaluate the usefulness of biological parameters in fetal blood as prognostic factors in relation with ultrasound features.

Materials and methods

All proven CMV infections referred to our centres between 2000 and 2006 were reviewed. In all cases, CMV fetal infection was proven to be the result of maternal primary infection as established by changes documented in maternal serological findings.24–26 We recently reported the results of a preliminary study with valaciclovir 8 g/day administered during pregnancy when fetuses were proven to be infected by CMV, and none of the cases treated with valaciclovir was included in this study.2 Women were referred to the two fetal medicine units (i) when primary infection was diagnosed based on serology screening either randomly performed (ii) when following nonspecific maternal symptoms or (iii) when abnormalities evocative of CMV fetal infection were diagnosed at ultrasound examination. Amniocentesis was performed in all cases after extensive counselling. CMV fetal infection was established by positive amplification of the viral genome by polymerase chain reaction (PCR) in the amniotic fluid.20 Congenital infection was considered certain when the viral genome was also detected either in fetal tissues after therapeutic abortion and stillbirth or in the urine samples of the newborns within the first 2 weeks of life using PCR.27

Once the diagnosis was made, serial targeted ultrasound examination was performed every fortnight. After appropriate counselling, two options were offered consisting of either cordocentesis for FBS in case of fetal infection or noninterventional management with serial ultrasound examination. FBS was performed under continuous ultrasound guidance, and fetal blood was analysed for platelet count, gamma-glutamyl transpeptidases (GGT), aminotransferases and total immunoglobulin M (IgM) plasma levels, as well as measurement of CMV DNA viral load. Decisions to continue or terminate the pregnancy were made by the women after extensive information on the fetal status. Women’s request for termination of pregnancy (TOP) was accepted according to the French law, whenever adverse outcome could be reasonably expected in relation to the severity of fetal infection. Whenever TOP was performed, postmortem macroscopic and microscopic examinations as well as virological testings were performed, and the severity of the disease was confirmed based on the severity of brain involvement and the degree of certainty that clinical neurological abnormalities would have developed if pregnancy had continued. For continuing pregnancies, a good outcome was defined as a newborn with normal development at the age of 6 months, irrespective of the results of transfontanellar ultrasound examination. Some infants who presented with minor abnormalities on transfontanellar scan, including isolated subependymal cysts, brain calcifications and echogenic vessels in the thalami, were classified in the ‘good outcome’ subgroup if they did not show any developmental delay, sensorineural impairment or mental restriction at 6 months of age. A poor outcome was defined by an abnormal clinical examination at the age of 6 months, including sensorineural hearing loss or TOP for progressive fetal abnormalities with severe disease confirmed at postmortem examination.

Fetal viraemia was diagnosed by amplification of the viral genome in fetal blood by real-time PCR with a threshold of 250 copies/ml to define positivity.20 Platelet count was evaluated in relation to gestational age. Thrombocytopenia was defined as a platelet count threshold of 120 000/mm3 and classified as severe (<70 000/mm3) or mild (70–120 000/mm3).28 Elevated gamma-glutamyl transpeptidases (GGT), aspartate aminotransferase and alanine aminotransferase (ALT) were defined based on normal ranges related to gestational age, as previously reported.29

Qualitative data were compared by means of chi-square test or Fisher’s exact test as appropriate. Student’s t test and Mann–Whitney test were used for parametric and nonparametric variables or unequal SDs, respectively. In case of possible associations, a robust logistic regression model with Huber–White sandwich estimator of variance being used in place of the traditional calculation was used to investigate the association between outcome and each of the possible explanatory parameter. Significant predictive parameters based on univariate analysis were included in the multivariate logistic regression model. Interaction between parameters was tested in the multivariate logistic regression model. All tests were two-tailed, and a P value <0.05 was considered statistically significant.


A total of 73 cases were included in this study. Maternal primary infection occurred during the periconceptional period, first and second trimester of pregnancy in 3, 56 and 6 cases, respectively. It remained undetermined in eight cases.

The outcome of 73 cases with proven CMV congenital infection is shown in Table 1. Overall, among the 73 cases analysed, a poor outcome was therefore observed in 35 cases (47.9%). Thirty-nine pregnancies continued up to term, women elected to terminate the pregnancy in 32 cases and intrauterine fetal deaths (IUFD) occurred in 2 cases. IUFD was a complication of FBS at 21 weeks of gestation in one case that showed hydrops, thrombocytopenia (85 000 platelets/mm3) and high viraemia (83 000 copies/ml). In the second case, IUFD occurred 6 weeks after the FBS performed at 23 weeks of gestation showing normal platelet count, no hepatitis and positive viraemia at 250 copies/ml. Autopsy was declined. Of 39 pregnancies that continued up to term, the outcome was considered to be good in 36 pregnancies. Among these, there were 28/36 cases without any imaging or developmental abnormality and 8/36 cases with isolated lenticulostriatal vasculitis at transfontanellar ultrasound examination without any evidence of developmental delay at 6 months of age. In 3/39 cases, the outcome was considered to be poor. In one case, unilateral sensorineural hearing loss (SNHL) was diagnosed at birth. In the second case, axial hypotonia, ventricular dilatation and multiple cerebral calcifications were observed at birth. In the last case, spontaneous preterm delivery occurred and the newborn died in the neonatal period from the prematurity-related complications, including respiratory distress, enterocolitis as well as diffuse CMV-related encephalitis revealed at autopsy. Major histological abnormalities related to CMV infection were observed in the fetal brain in 30 of the 32 cases that underwent TOP. However, only minor abnormalities (focal encephalitis) were observed in another two cases.

Table 1.  Outcome of the 73 cases of congenital CMV infections
 Poor outcomeGood outcome
TOP (n = 32) (44%)Major histological abnormalities related to CMV infection (n = 30)Mild histological findings at 23 and 27 weeks (n = 2)
IUFD (n = 2) (3%)Case 1 (following FBS): hydrops fetalis, thrombocytopenia, positive fetal viraemia; Case 2 (6 weeks after FBS): no biological abnormality, no ultrasound findings, autopsy not performed 
Continuing pregnancies (n = 39) (53%)Case 1: Persistant axial hypotonia, disseminated intracerebral calcifications and ventricular dilatation; Case 2: Unilateral sensorineural hearing loss; Case 3: Neonatal death from severe prematurity and CMV encephalitis (n = 3)No ultrasound features, no biological abnormalities at follow up (n = 36)

Ultrasound findings were available in all cases. The prognostic value of ultrasound abnormalities is summarised in Table 2. Ultrasound features of infection were observed in 38 of the 73 cases (52%). Among these, the central nervous system (CNS) was involved in 27 (37%) cases mainly showing hydrocephaly and brain calcifications. Extracerebral abnormalities were observed in 28 cases, and these were isolated in 11 cases. The most frequent extracerebral ultrasound abnormalities were hepatomegaly and hyperechogenic bowel. Hydrocephaly and brain calcifications were the most common CNS abnormalities (Table 3). A poor outcome was observed in 30 (79%) and 5 (14%) cases in fetuses with and without ultrasound abnormalities, respectively. This difference was highly significant (P < 10−4). Moreover, a poor outcome was observed in 24/27 (89%) and 11/46 (24%) cases in fetuses with and without cerebral ultrasound abnormalities, respectively (Table 5).

Table 2.  Analysis of the association of biochemical, haematological and ultrasound parameters with outcome
Parameters Poor outcomeGood outcomeTotalP values
  • AST, aspartate aminotransferase; NS, nonsignificant.

  • *

    Related to gestational age.28,29

ViraemiaPositive2228500.5 (NS)
Platelet countAbnormal24933<10−4
Mean (SD)111.103 (65.103)175.103 (45.103)<10−4
GGTAbnormal*1919380.9 (NS)
ASTAbnormal*89170.286 (NS)
ALTAbnormal*2240.683 (NS)
IgMPositive1814320.224 (NS)
Any ultrasound abnormalityPresent30838<10−4
Isolated noncerebral ultrasound abnormalityPresent65110.01
Cerebral ultrasound abnormalityPresent24327<10−4
Table 3.  Prenatal ultrasound findings in fetuses with CMV infection
Findingsn (%)
  • IUGR, intrauterine growth restriction.

  • *

    This was associated with cerebral abnormalities in 17 cases.

Noncerebral ultrasound abnormalities* (28 cases)Hyperechogenic bowel19 (26)
Hepatomegaly10 (14)
IUGR7 (10)
Oligohydramnios5 (7)
Ascite3 (4)
Liver calcifications2 (3)
Enlarged placenta2 (3)
Polyhydramnios1 (1)
Pericardial effusion1 (1)
Cardiomyopathy1 (1)
Hydrops1 (1)
Right club foot1 (1)
Cerebral ultrasound abnormalities (27 cases)Brain calcifications13 (18)
Hydrocephaly13 (18)
Microcephaly7 (10)
Germinative cysts5 (7)
Ventricular dilatation2 (3)
Holoprosencephaly1 (1)
Polymicrogyria1 (1)
Cystic occipital lesions1 (1)
Temporal cysts1 (1)
Table 5.  Proportion of cases with normal or abnormal outcome depending on ultrasound and FBS
 Abnormal cerebral ultrasound findings (n/n)Abnormal noncerebral ultrasound findings (n/n)Only noncerebral findings (n/n)Normal ultrasound (n/n)Total (n/n)
  1. (n/n), abnormal outcome/normal outcome.

 27 (24/3)28 (23/5)11 (6/5)35 (5/30) 
Platelet count: normal12 (10/2)10 (7/3)3 (0/3)30 (3/27)45 (13/32)
Low platelet count13 (12/1)16 (14/2)8 (6/2)3 (2/1)24 (20/4)
Abnormal GGT13 (11/2)14 (10/4)7 (3/4)14 (3/11)35 (17/18)
Normal GGT10 (9/1)9 (8/1)2 (1/1)15 (2/13)27 (12/15)
Total27 (24/3)28 (23/5)11 (6/5)35 (5/30) 

The mean gestational age at cordocentesis was 27.4 weeks (range = 19–38 weeks). Table 2 summarises the prognostic value of fetal haematological and biochemical parameters. Platelet count was strongly associated with outcome when considered either as a qualitative (normal versus abnormal platelet count) or as a quantitative variable. None of the other haematological and biochemical parameters was significantly associated with outcome.

Based on univariate logistic regression, odds ratio for a poor outcome were 1.24, 7.2, 22.5 and 25.5 for each 10 000/mm3 decrease in platelet count, the presence of noncerebral, any ultrasound and cerebral ultrasound abnormalities, respectively (Table 4). Based on a multivariate model, these predictors remained significantly associated with a poor outcome (Table 5). Using this multivariate model, the area under the receiver operating characteristic curve (Figure 1) for predicting a poor outcome would be 0.9. With a cutoff value at 0.5 for the multivariate logistic regression, our predicting model for a poor outcome would have a sensitivity of 81.8%, specificity of 86.1%, positive predictive value of 84.4% and negative predictive value of 83.8%.

Table 4.  Prognostic value of ultrasound features and biological parameters (logistic regression)
ParametersOR and adjusted OR for poor outcome95% CIP values
Univariate analysis
Thrombocytopenia (for each 10 000/mm3 decrease)1.241.10–1.39<10−4
Noncerebral ultrasound anomaly (if present)7.21.5–33.40.01
Any ultrasound anomaly (if present)22.56.5–73.4<10−4
Cerebral ultrasound anomaly (if present)25.56.4–101.9<10−4
Multivariate analysis
Thrombocytopenia (for each 10 000/mm3 decrease)1.131.06–1.20<10−4
Ultrasound findingsNo finding1
Noncerebral ultrasound anomaly4.41.3–150.02
Cerebral ultrasound anomaly40.68.0–206.9<10−4
Figure 1.

Receiver operating characteristic curve (ROC) obtained with the multivariate regression model (a 10 000/mm3 decrease in platelet count, the presence of noncerebral ultrasound abnormality, any ultrasound abnormality and a cerebral ultrasound abnormality) for the prediction of a poor outcome. With a cutoff value at 0.5 for the multivariate logistic regression, sensitivity, specificity, positive predictive value and negative predictive value would be 81.8, 86.1, 84.4 and 83.8%, respectively. Area under the curve = 0.9091.


The lack of prognostic factors in intrauterine CMV infection has led most developed countries to abstain from offering maternal serology screening during pregnancy. This study demonstrates that the presence of abnormalities at prenatal ultrasound examination and thrombocytopenia appears to be strong and independent prenatal predictors for a poor outcome.

Prenatal assessment should, however, account for the fact that fetal ultrasound anomalies can sometimes be revealed only late in the pregnancy, and they may change or even disappear with time.30–32 This justifies serial targeted ultrasound examination by trained operators because CMV may induce subtle changes. Visualisation of any ultrasound abnormality is often the basis for accepting women’s request for TOP.

In our series, among 33 liveborn-infected infants, 17 had abnormal ultrasound features and were born with severe inclusion disease (CID), but 2/16 with normal ultrasound examination also had CID, while 14 were asymptomatic. Several reports have also underlined a good outcome, despite the presence of ultrasound abnormalities.20,33–35

The need for a more accurate evaluation has led to investigate postnatal findings in symptomatic neonates to be applied to infected fetuses.18 This includes thrombocytopenia and hepatitis. Fetal blood can also be examined for the presence and quantification of human CMV-specific IgM and of viral load as well as for the assessment of biochemical and haematological parameters.21,22,32,35 Fetal blood examination is not useful to make the diagnosis of fetal infection and carries a risk of fetal loss that has to be balanced with the need for prognostic evaluation.36,37 Revello and Gerna12 reported a higher median level of human CMV-specific IgM in the blood of symptomatic newborns. Enders et al.38 reported that symptomatic disease occurred in 29/31 fetuses with specific IgM and in only 3/6 fetuses without detectable IgM. In our study, the level of specific IgM was not correlated with the outcome. Rivera et al.39 have reported that neonatal thrombocytopenia (platelet count <100 000/mm3) and high levels of ALT (>80 UI/ml) were associated with the presence of symptoms at birth. Few studies have reported the prognostic value of these factors during fetal life. Liesnard et al.35 have described nine fetuses undergoing TOP with thrombocytopenia and four with ultrasound abnormalities. Furthermore, 2/29 infected newborns had thrombocytopenia and both also showed ultrasound abnormalities in utero.35 In another TOP series, four of eight fetuses had thrombocytopenia and three of these also showed ultrasound abnormalities.40 Our study shows that fetal thrombocytopenia is an independent prognostic factor of a poor outcome. Furthermore, we have observed that the prognosis worsens as platelet count decreases.

In conclusion, our results suggest that despite a procedure-related loss rate around 1–3%,36,37,41 FBS for fetal platelet count determination could be a useful adjunct to prenatal ultrasound examination in fetuses that have been diagnosed with CMV infection. This could be particularly relevant when there is no cerebral anomaly at ultrasound examination.