We thank Leyder et al for their contribution that addresses the important and challenging issue of predicting outcome when cytomegalovirus infection is diagnosed prenatally. Their group in Brussels has a long-standing experience in this field. The study confirms findings by us and others that the time of maternal infection is a prognostic factor, because all cases with a poor outcome had an infection before 15 weeks of gestation (2 cases with unknown timing). The other major factor to be associated with poor outcome is the presence of ultrasound abnormalities.

Leyder et al conclude that, of 38 fetuses with normal ultrasound results, more than one- of them half had cytomegalovirus-related anomalies at clinical follow up or at autopsy in case of pregnancy termination, which led them to state that prenatal ultrasound findings are correlated poorly with postnatal outcomes.

This conclusion may be overstated for several reasons, in addition to the number lost to follow up. First, the paper fails to mention when the ultrasound examinations were performed. This is important because it has been shown that ultrasound signs may take several weeks or months to appear. Second, the autopsy findings must be interpreted with caution. It is impossible to determine whether microscopic lesions would have resulted in neurologic impairment had the child lived, but the presence of cytomegalovirus inclusions in the placenta, lungs, or abdomen does not lead necessarily to neurologic impairment. Furthermore, since microscopic lesions may precede visible abnormalities by many weeks, ongoing surveillance with serial ultrasound examinations would have been likely to detect the most severe evolving lesions, microcephaly for instance. Third, regarding the liveborn children, of 23 children with normal prenatal ultrasound scans, 1 child had had mild developmental delay, and 1 child had mild neurologic sequelae, thus a negative predictive value of 91% to exclude neurologic problems. Of course, ultrasound scans cannot detect hearing loss, which much be discussed with the couple.

Thus, the negative predictive value is underestimated in this article. Furthermore, more complete evaluation may be considered with fetal blood sampling for platelet count and magnetic resonance imaging. This is important with regards to the high proportion of pregnancies that were terminated, despite normal ultrasound findings. Lack of confidence in favorable prognostic elements and serial follow-up evaluations can lead to distress leading to termination of pregnancy.

Thanks to the better knowledge of the natural history of the disease, we now can establish with good reliability which fetuses will go on to be asymptomatic neonates with a low risk of neurologic sequelae. Of course, the prediction is not perfect, and we cannot exclude the occurrence of hearing loss. We strongly agree with Leyder et al that further longitudinal studies are required. Management of these cases must be performed with experts on fetal ultrasound scanning and cytomegalovirus disease.