Objective. To determine whether newborn cranial computed tomographic (CT) scan abnormalities predict an adverse neurodevelopmental outcome in children with symptomatic congenital cytomegalovirus (CMV) infection and to examine the association between clinical findings at birth and imaging abnormalities.
Methods. The data from 56 children with symptomatic congenital CMV infection who underwent cranial CT scans as newborns and were enrolled in a long-term follow-up study were analyzed. The incidence of sequelae was compared between the groups of children with normal and abnormal imaging studies. The relationship between CT scan results and other newborn findings was also examined.
Results. Abnormal CT scans were noted in 70% of subjects; intracerebral calcification was the most frequent finding. Most of the children with an abnormal newborn CT scan (90%) developed at least one sequela, compared with 29% of those with a normal study. Only 1 child with a normal CT scan had an IQ <70, in contrast to 59% of those with imaging abnormalities. In addition, almost half of the children with CT abnormalities had an IQ <50 compared with none of those with a normal CT scan. Newborn CT abnormalities were also associated with an abnormal hearing screen at birth and hearing loss on follow-up. None of the neonatal neurologic findings were predictive of an abnormal CT scan.
Conclusion. In neonates with symptomatic congenital CMV infection, a cranial CT scan is a good predictor of an adverse neurodevelopmental outcome. In addition, newborn clinical and laboratory findings did not predict neuroradiographic abnormalities in neonates with symptomatic congenital CMV infection.
Congenital cytomegalovirus (CMV) infection is a leading cause of brain disease and hearing loss in children.1,2 An estimated 40 000 infants are born each year with congenital CMV infection, and up to 8000 of these children will develop permanent central nervous system sequelae.1-6 The sequelae include sensorineural hearing loss, mental retardation, cerebral palsy, seizures, and visual defects.1-6 Most of the infants with intrauterine CMV infection are without apparent clinical manifestations at birth (asymptomatic), and only ∼10% of the neonates exhibit findings suggestive of congenital infection (symptomatic infection).1,7,8 Although it is clear that the majority of infants with symptomatic congenital CMV infection develop perceptual or neurologic deficits, the predictors of an adverse long-term outcome have not been identified.8-10
Intracerebral calcification is the most common abnormality seen on neuroradiologic imaging of infants with congenital CMV infection.10-12 An association between periventricular radiolucencies or calcifications in the newborn period and an adverse early developmental performance was demonstrated in children with asymptomatic congenital CMV infection.13 The same group of investigators also reported that the presence of early computed tomographic (CT) scan abnormalities was the only factor that predicted the occurrence of hearing loss in children with asymptomatic congenital CMV infection.13 Although it is widely believed that CT scan abnormalities such as intracerebral calcifications in the newborn period are associated with an adverse outcome, this issue has not been examined in long-term follow-up studies of children with symptomatic congenital CMV infection. The purposes of our study were (1) to test the assumption that an abnormal CT scan early in life is predictive of an adverse developmental outcome in children with symptomatic congenital CMV infection, and (2) to determine whether any of the clinical or laboratory findings in the newborn period predict an abnormal imaging study.
MATERIALS AND METHODS
Fifty-eight children with symptomatic congenital CMV infection who were born between 1983 and 1994 and underwent noncontrast cranial CT imaging during the first month of life were identified. The majority of children underwent CT imaging at the Children's Hospital of Alabama (38/58, 66%), and for the rest of the infants, the films were reviewed by a pediatric radiologist at our institution. Two of the infants died from complications related to congenital CMV infection during the first month of life and were excluded from the data analysis, because follow-up data was not available. Informed consent was obtained from parents or legal guardians of the study children. Infants were categorized as symptomatic if any of the following were observed at birth: petechiae, hepatosplenomegaly, microcephaly, thrombocytopenia, or jaundice with conjugated hyperbilirubinemia.7Intrauterine CMV infection was documented by the isolation of virus from urine or saliva during the first 2 weeks of life.14,15The clinical and laboratory findings during the newborn period and the results of the cranial CT scans reviewed by a pediatric radiologist were recorded on standardized case report forms and compiled in SAS datasets (SAS Institute, Cary, NC). The study subjects were followed in a special clinic; serial assessments of neurologic, developmental, audiologic, and visual status were performed according to the standard methods described previously.3 The sequelae data were accumulated longitudinally and compiled in an SAS dataset.
The original CT scans could only be retrieved in a subset of 20 children, and those scans were evaluated independently by two pediatric radiologists blinded to the results of the initial review and the patient's long-term neurologic outcome. The data from this independent review were analyzed for interobserver agreement, and a kappa statistic (k value), which adjusts for chance agreement, was obtained using the Epi-Info statistical software.16,17k values of >.75, .75-.45, and <.45 were considered to represent excellent, good, and poor agreement, respectively, of the results beyond chance between the two observers.18
We sought next to determine whether the previously recorded CT scan results in the newborn database from the subpopulation of 20 children concurred with the independent review by the radiologists. The agreement between the review by the independent radiologists and the results from the newborn database was shown to be excellent (k ≥ .75); therefore, the results of CT scans from the 56 children in the database were included in the final analysis to determine the association between CT abnormalities and the long-term outcome. The newborn CT findings and clinical measures of outcome were compared by univariate analyses. Odds ratios (OR), together with the exact 95% confidence intervals (CI), were determined.16 In addition, the relationship between CT abnormalities and other newborn findings was ascertained by comparing frequencies using χ2 or Fisher's exact test where appropriate.
The results of the blinded review of the cranial CT scans from a subset of 20 study subjects by two pediatric radiologists were compared to determine the interobserver agreement using a kappa statistic. As can be seen in Table 1, there was good agreement between the two observers in categorizing CT scans as abnormal, the presence of intracranial calcifications, white matter abnormalities, ventricular dilatation, and cortical atrophy. However, there was considerable disagreement in the detection of migration abnormalities (k = .15). Similar to the interobserver comparison, excellent agreement between the findings of the radiologists and those in the newborn database was demonstrated in categorizing CT scans as abnormal and in identifying intracerebral calcifications with a k value >.75 (Table 1). The imaging data from all 56 children contained in the newborn database were included in the subsequent analysis because of an excellent concordance between the results of the independent review by radiologists and findings in the newborn database.
Characteristics of the group of children with a normal CT scan and the group with imaging abnormalities are shown in Table 2. The majority of children with a normal CT scan were black (53%) compared with 26% of those with an abnormal study. A significantly higher proportion of children with imaging abnormalities (82%) were referred to us for evaluation and enrollment in an ongoing longitudinal study. In contrast, ∼50% of the infants with normal scans (53%) were detected by newborn virologic screening. Other demographic features, including sex, gestational age, birth weight, insurance status, and the source of maternal prenatal care, were not significantly different between the two groups. The median duration follow-up was not significantly different between the referral group (median: 35 months; range: 1 to 143.5 months) and the group identified by newborn screening (median: 29.2 months; range .7 to 98 months). Similarly, the age at initial evaluation was not significantly different between the two groups (median: 2 months vs 2.4 months).
The frequency of the various newborn CT abnormalities in the study population is shown in Table 3. An abnormal CT scan was noted in 70% (39/56) of subjects; intracerebral calcifications were the most frequent finding and were seen in 77% of those with an abnormal scan. Additional abnormalities including ventricular dilatation and cortical atrophy were seen in 10% of the study subjects. Thus, approximately one third of neonates with symptomatic congenital CMV infection who underwent CT imaging of the brain had a normal imaging study. The sequelae data from the long-term follow-up of the study population are shown in Table 4. The results of hearing evaluations were available for all 56 children, whereas psychometric evaluations were performed in 36 subjects. Sensorineural hearing loss was seen in 59% of the children, cerebral palsy in 49%, mental retardation (IQ <70) in 47%, chorioretinitis in 12%, and seizures in 11%. Severe mental retardation (IQ <50) was observed in 36% of the children with psychometric evaluations. At least one neurodevelopmental sequela was noted in 71% of the study population.
The incidence of neurologic sequelae was compared between the group of children with an abnormal newborn CT scan and the group with normal imaging studies (Table 5). Most of the children (90%) with an abnormal cranial CT scan developed at least one sequela, whereas only 29% of children with a normal study had sequelae (OR = 16.8; P < .001). Only one child in the group with normal CT scans was noted to have mental retardation (IQ <70), in contrast to 59% of the children with an abnormal newborn cranial CT scan (OR = 11.6; P = .015). Furthermore, none of the children with a normal CT scan were noted have severe mental retardation, whereas almost half of the children with abnormalities on CT imaging had IQ <50. When the occurrence of any neurodevelopmental sequela including mental retardation, cerebral palsy, and/or seizures was compared, 77% of children with abnormal CT scan findings developed at least one psychomotor sequela, compared with only 12% of children with a normal CT scan (OR = 24; P < .001). Furthermore, the presence of intracranial calcifications was more likely to be associated with severe mental retardation than the presence of other abnormal CT scan findings. A significantly higher proportion of the children with abnormalities on CT imaging (72%) developed sensorineural hearing loss compared with 29% of children with normal scans (OR = 5.6; P < .01). The wide range in the CI values (Table 5) is indicative of the small sample size, preventing a more precise estimation of the true risk of neurologic sequelae in the study population.
To determine whether clinical and laboratory findings at birth could predict an abnormal neuroimaging study in symptomatic congenital CMV infection, the results of the CT scans were compared with other newborn findings (Table 6). There was no association between findings such as jaundice, hepatosplenomegaly, and purpura and the presence of abnormalities on the CT scan. However, the presence of petechiae in the newborn period was correlated with imaging abnormalities (P = .04). None of the clinical neurologic findings including microcephaly, seizures, lethargy, or poor suck predicted an abnormal newborn CT scan (P = .29). Findings such as chorioretinitis and optic atrophy on eye examination in the newborn period were associated with imaging abnormalities, although this correlation did not reach statistical significance (P = .08). Except for thrombocytopenia (<100 000/mm3), none of the other laboratory findings including elevated aspartate aminotransferase (>80), conjugated hyperbilirubinemia (direct bilirubin >2.0 mg/dL), and elevated cerebrospinal fluid protein (>120 mg/dL) were associated with imaging abnormalities. There was no correlation between severe thrombocytopenia (<50 000/mm3) and abnormal CT scan findings. Within the group of infants with an abnormal CT scan, 68% failed a newborn hearing screen compared with 31% of children with a normal imaging study (P = .03).
Our findings demonstrate that cranial CT scan abnormalities during the newborn period are predictive of an adverse neurologic and developmental outcome in children with symptomatic congenital CMV infection. Abnormal CT scan findings were observed in more than two thirds of the neonates (70%); intracranial calcifications, the most frequent finding, were seen in 77% of those with imaging abnormalities. More than half of the children with an abnormal newborn CT scan (59%) developed mental retardation (IQ <70) in contrast to only 1 child who had a normal study. More striking was the association with severe mental retardation (IQ <50), because almost half of the children with imaging abnormalities had an IQ <50, whereas none of the patients with a normal CT scan exhibited severe mental retardation. Only 12% of the infants with a normal CT scan developed psychomotor sequelae, in contrast to 77% of those with CT abnormalities, suggesting that a normal neuroimaging study at birth in symptomatic congenital CMV infection predicts a good long-term neurologic outcome. In addition, the presence of intracerebral calcifications was associated with the development of severe mental retardation more strongly than other CT abnormalities. A significant association between an abnormal newborn CT scan and sensorineural hearing loss was also documented. Although the calculated OR values for the occurrence of various neurologic sequelae were significant in children with an abnormal newborn CT scan (5.3 to 24.2), a precise estimate of risk was not possible because of the small sample size. However, the value of newborn cranial CT scan abnormalities for predicting central nervous system sequelae was demonstrated by the observation that the lower limit of the CI values for all the sequelae examined except seizures was >1.0.
Because there was a significant association between newborn CT scan findings and the incidence of sequelae on long-term follow-up, we wanted to determine whether any of the newborn clinical and laboratory findings were associated with an abnormal imaging study. Such an association between the clinical findings and an abnormal CT scan in the newborn period could permit selection of infants for obtaining neuroradiologic studies at birth. In a previous study of 32 symptomatic children, the presence of chorioretinitis and microcephaly at birth was significantly associated with mental retardation.10 The results of our study also documented an association between an abnormal ophthalmologic examination and the presence of CT scan abnormalities in the newborn period, although this relationship did not reach statistical significance. However, there was no correlation between clinical neurologic findings such as microcephaly, seizures, lethargy, and poor suck and an abnormal CT scan. A lack of association between a normal neurologic examination at birth and a favorable developmental outcome in children with symptomatic congenital CMV infection was also shown in a previous report from our laboratory.9 The results of the present study extended these findings and showed that it was not possible to predict imaging abnormalities in infants with symptomatic congenital CMV infection using the clinical and other laboratory findings at birth.
A blinded review of the newborn cranial CT scans by the two pediatric radiologists could be performed only in a subset of 20 study subjects in whom the original studies were available for the review. We were unable to determine the value of CT scan findings in the prediction of long-term outcome from this independent review, because most of the neonates in this subset (17/20) had abnormal CT scans. The inclusion of the data from all 56 study subjects with symptomatic congenital CMV infection in the subsequent analysis could be justified because of an excellent concordance (k >.75) between the blinded review by the radiologists and CT summaries in the newborn database. Because <50% of the children who were followed at our institution with symptomatic congenital CMV infection during the study period underwent CT scans as newborns, it is possible that the imaging studies were obtained in infants with a more severe involvement of the central nervous system. However, the absence of an association between the newborn findings including abnormal neurologic examination and CT scan abnormalities indicates that the results of our study could not be attributed to such a selection bias. Because the majority of the study children were referred to us for evaluation (rather than detected by newborn virologic screening), it could also be argued that these infants had more severe intrauterine infection and clear evidence of congenital infection in the newborn period. However, almost one third of the study subjects had normal CT scans and more than two thirds of subjects with normal studies at birth escaped long-term central nervous system sequelae, suggesting that the referral bias could not account for our findings. Furthermore, the results of our study supported the findings reported by Williamson et al.13 documenting a strong association of the imaging abnormalities with a failed neonatal hearing screen and hearing loss on long-term follow-up in children with asymptomatic congenital CMV infection.
Congenital CMV infection is the most frequent congenital infection with ∼40 000 infected infants born each year in the United States; ∼4000 of these children are symptomatic at birth.1-3Although the majority of children with symptomatic congenital CMV infection develop permanent central nervous system sequelae, the results of our recent studies of children identified by newborn virologic screening suggest that ∼40% of these children escape complications and develop normally (unpublished observations). Thus, the association between newborn neuroradiographic abnormalities and an adverse neurodevelopmental outcome is an important finding that could help in the identification of infants at risk for the sequelae after symptomatic congenital CMV infection. Early recognition of these high-risk newborns should target them for a close follow-up and early intervention. This information, as well as the finding that a normal newborn cranial CT scan is a good predictor of a normal developmental outcome, provides valuable prognostic information for parents of infants with symptomatic congenital CMV infection. All newborns with symptomatic congenital CMV infection should be considered for neuroradiologic imaging in an attempt to identify those at risk for neurodevelopmental sequelae. Because none of the newborn neurologic findings predicted imaging abnormalities in our study population, it may not be possible to restrict neuroradiographic studies to those infants with clinical neurologic abnormalities.
This work was supported in part by grants from the National Institutes of Health, National Institute of Deafness and Other Communication Disorders (K01 DC00079), National Institute of Child Health and Human Development (P01 HD10699), the Research Institute of the Children's Hospital of Alabama, and the Joyce Family Foundation.
- CMV =
- cytomegalovirus •
- CT =
- computed tomographic (scan) •
- OR =
- odds ratio •
- CI =
- confidence interval
- ↵Stagno S. Cytomegalovirus. In: Remington JS, Klein JO, eds. Infectious Diseases of the Fetus and Newborn Infant. 4th ed. Philadelphia, PA: WB Saunders; 1995;312–353
- Hanshaw JB
- Bale JF,
- Blackman JA,
- Sato Y
- Pass RF,
- Stagno S,
- Myers GJ,
- Alford CA
- Williamson WD,
- Demmler GJ,
- Percy AK,
- Catlin FI
- Boppana SB,
- Smith R,
- Stagno S,
- Britt WJ
- Balcarek KB,
- Warren W,
- Smith RJ,
- Lyon MD,
- Pass RF
- ↵Dean AG, Dean JA, Coulombier D, et al. Epi Info, Version 6. Atlanta, GA: Centers for Disease Control and Prevention; 1994
- ↵Rosner B. Fundamentals of Biostatistics. 2nd ed. Boston, MA: Duxbury Press; 1986
- Copyright © 1997 American Academy of Pediatrics