Introduction
Congenital cytomegalovirus (CMV) infection affects 0.7% of live births and is the leading cause of congenital neurological disease of infectious origin. Among all neonates positive for infection on screening, 20% eventually have neurodevelopmental impairment with permanent sequelae. Around 10% of infected neonates are symptomatic at birth; their risk of sequelae reaches 58%, including sensorineural hearing loss or cognitive or motor defects. The risk of sequelae in newborns who were asymptomatic at birth is around 13%, mainly due to progressive hearing loss.
An infected fetus’s risk of symptoms at birth is assessed by interpreting the results of both prenatal imaging and laboratory tests. Fetal CMV disease is progressive: early symptoms of systemic infection can be expressed as extracerebral findings at prenatal ultrasound; fetal brain involvement usually does not show until several weeks later. Severe brain lesions seen on prenatal ultrasound predict a dismal outcome. This leaves a window of opportunity for treatment of symptomatic fetuses without brain involvement.
Vaccination is not available and no prenatal treatment of congenital CMV has yet been validated. The use of CMV-specific hyperimmune globulin to prevent transmission from mother to fetus has produced conflicting results. Neonatal antiviral treatment with either ganciclovir or valganciclovir improves auditory and neurological outcomes in symptomatic newborns, but these drugs, highly genotoxic in vitro, are not approved in pregnancy. Although valacyclovir is less effective than ganciclovir against CMV in vitro, high-dosage valacyclovir has proven clinically efficient to prevent CMV disease in transplant recipients. The mechanism of acyclovir’s anti-CMV activity in clinical settings remains unexplained. Valacyclovir also has the best safety profile of the anti-CMV drugs. Neither cell transformation nor increased risk of neoplasia has been reported in vitro, and no increased risk of birth defects has been detected in the offspring of thousands of women exposed during pregnancy. Finally, valacyclovir is well tolerated with rare side effects. In a pilot study, we found that oral administration of high-dosage valacyclovir to mothers significantly decreased viral load and produced therapeutic concentrations in fetal blood with a mean fetal blood plasma concentration of >17 μmol/L. These results suggested the value of a clinical trial to investigate this therapeutic option further. We failed to complete a randomized controlled trial comparing prenatal treatment with valacyclovir against placebo in moderately symptomatic infected fetuses due to failure to recruit (Cymeval NCT01037712 ). In this open-label phase II trial with 1 arm we show that high-dosage valacyclovir given in pregnancy is safe and appears effective for improving the outcome of moderately symptomatic fetuses.
Materials and Methods
Patients
Eligible women were pregnant with an infected fetus identified by a positive CMV polymerase chain reaction assay in amniotic fluid, sampled by amniocentesis >21 weeks, together with the presence of ≥1 extracerebral ultrasound features compatible with CMV infection and/or 1 isolated cerebral abnormality and/or 1 of the following laboratory findings in fetal blood: fetal platelet count <100,000/mm 3 or CMV DNA viral load >3000 copies/mL ( Table 1 ). The presence of severe ultrasound brain abnormalities ( Table 2 ) and the absence of any ultrasound feature of infection or laboratory abnormality in fetal blood were exclusion criteria. The detailed eligibility criteria are listed in the supplementary Appendix . Fetal blood sampling by cordocentesis under ultrasound guidance was offered to all participants to evaluate fetal platelets and viral DNA load to help refine the fetal prognosis. Cordocentesis was not, however, required for study eligibility, as this invasive procedure was not the standard of care in all participating centers.
At least 1 extracerebral abnormality compatible with fetal CMV infection | |
Fetal growth restriction | |
Abnormal amniotic fluid volume | |
Ascites and/or pleural effusion | |
Skin edema | |
Hydrops | |
Placentomegaly >40 mm | |
Hyperechogenic bowel a | |
Hepatomegaly >40 mm | |
Splenomegaly >30 mm | |
Liver calcifications | |
And/or 1 isolated cerebral abnormality | |
Moderate isolated ventriculomegaly (<15 mm) | |
Isolated cerebral calcification | |
Isolated intraventricular adhesion | |
Vasculopathy of lenticulostriate vessels | |
And/or laboratory findings of generalized CMV infection in fetal blood | |
Fetal viremia >3000 copies/mL | |
Fetal platelet count <100,000/mm 3 |
a Since diagnosis of hyperechogenic bowel can be subjective and associated with high interoberver and intraobserver variability, diagnosis of hyperechogenic bowel was only considered for grade-2 or grade-3 hyperechogenic bowel ; this semiquantitative analysis was chosen to limit subjectivity sometimes associated with ultrasound findings.
Presence of at least 1 severe cerebral ultrasound abnormality among following: | |
Ventriculomegaly ≥15 mm | |
Periventricular hyperechogenicity | |
Hydrocephaly | |
Microcephaly <3 SD | |
Mega-cisterna magna >10 mm | |
Vermian hypoplasia | |
Porencephaly | |
Lissencephaly | |
Periventricular cysts | |
Abnormal corpus callosum | |
Or absence of any ultrasound feature of infection or laboratory abnormality in fetal blood |
Procedures
Study design
The trial was a multicenter, open-label, phase II study with 1 arm, based on 1 of Simon’s optimal 2-stage designs. All participants received oral valacyclovir (2 g, 4 times a day, therefore 8 g daily). The medication was continued until delivery or 24 treatment weeks, whichever was sooner. The study drug was purchased from GlaxoSmithKline (Marly Le Roi, France), which had no other role in the study. Visits for clinical and ultrasound examinations, questions about clinical side effects (headache, nausea, neurologic effects listed in the product monograph), and adherence assessment by pill count were scheduled every 2 weeks until delivery. Maternal plasma levels of aspartate aminotransferase, alanine aminotransferase, and creatinine were assessed once a month during treatment. All newborns were examined between days 4 and 7 of life by a trained pediatrician according to a standardized clinical evaluation. Auditory brainstem responses, cranial ultrasound, fundoscopy, and laboratory tests (see criteria for primary endpoint in supplementary data ) were also performed.
The ethics committee of Poissy-Saint Germain Hospital approved the study (2011-001610-34). Participants gave written informed consent before inclusion. Study oversight was provided by an independent data and safety monitoring board (Clinical Research Unit, Cochin-Necker, Paris).
Study endpoints
The primary study endpoint was the proportion of asymptomatic neonates born to women treated with valacyclovir. An asymptomatic neonate was a neonate without growth restriction (that is, with birthweight ≥10th percentile), normal clinical examination, normal laboratory findings, no severe features of infection on cerebral imaging, normal funduscopic examination, and normal audiology findings (see the supplementary Appendix for details).
The secondary endpoints included adverse events related to the study medication and adherence to treatment. CMV DNA levels and platelet counts were compared in pretreatment fetal blood when available and cord blood at birth and in symptomatic and asymptomatic neonates. The effect of the duration of maternal treatment was also assessed.
Historical comparator group
Our systematic review on PubMed using 3 key words (“cytomegalovirus,” “congenital,” “ultrasound”) yielded 216 articles, but only 3 were written in English and included detailed tables that described both prenatal ultrasound findings and neonatal outcome and also included postmortem findings in termination of pregnancy cases. These 3 suitable studies included 724 pregnancies with a maternal CMV primary infection, 217 with an infected fetus. The review of the ultrasound symptoms of these 217 fetuses showed no ultrasound abnormalities in 142 of them and severe cerebral ultrasound abnormalities in 28. Ultrasound abnormalities matching our inclusion criteria were therefore seen in 47 cases, which formed the historical comparator group. Among them 20 (42.55%) neonates were born asymptomatic. Twenty fetuses were terminated and all underwent postmortem examination showing both macroscopic and microscopic evidence of brain damage in all cases. We therefore assumed that these fetuses would have been born symptomatic and classified then into the symptomatic group. See Table S1 for details in the supplementary Appendix .
Laboratory assays
CMV serology and viral load quantification were centralized and analyzed at the Necker Hospital virology laboratory. Maternal CMV primary infection was diagnosed by seroconversion or the concomitant presence of CMV-specific IgM antibodies and low IgG avidity. After extracting DNA from fetal whole blood, cord whole blood, and neonatal urine with the MagnaPure LC platform (Roche Diagnostic, Meylan, France), we performed quantitative polymerase chain reaction for CMV with the CMV-R gene kit (Argene, BioMerieux, Marcy-L’Etoile, France). The limit of detection was 446 copies (or 178 IU)/mL.
Statistical analysis
To estimate the sample size according to Simon optimal 2-stage design, we assumed that a proportion of asymptomatic neonates of <60% was not clinically relevant in relation to valacyclovir efficacy, while a proportion of >80% was deemed acceptable. With a type I error fixed at 0.05 and a power of 80%, we needed 11 infected fetuses for the first stage: if at least 8 cases in stage 1 had a good outcome (were asymptomatic), then 32 additional infected fetuses would be included in stage 2 for a total of 43. If at least 31 of the 43 cases were asymptomatic at birth, valacyclovir would be judged to have a positive effect.
Analyses of efficacy data were performed for all included fetuses (intention to treat). The proportion of asymptomatic neonates and its 95% confidence interval were calculated with an unbiased estimator. In the historical comparator group, we used a random effects model to calculate the overall proportion of asymptomatic neonates and a 95% confidence interval for all 47 cases from the 3 studies.
We used nonparametric tests for all secondary analyses.
Platelet counts at birth and in fetal blood were compared to the gestational age-specific reference range published by Meher-Homji et al. The average individual deviations of values at birth and in fetal blood from the published mean in noninfected fetuses were compared to 0. We also compared individual differences in platelet counts between fetal blood and birth to the slope of the reference mean in noninfected fetuses. Two-sided t tests were used for these analyses.
R software (Version 2.15; R Foundation for Statistical Computing, Vienna, Austria) was used for all statistical analyses and its “meta” package for the metaanalysis.
Materials and Methods
Patients
Eligible women were pregnant with an infected fetus identified by a positive CMV polymerase chain reaction assay in amniotic fluid, sampled by amniocentesis >21 weeks, together with the presence of ≥1 extracerebral ultrasound features compatible with CMV infection and/or 1 isolated cerebral abnormality and/or 1 of the following laboratory findings in fetal blood: fetal platelet count <100,000/mm 3 or CMV DNA viral load >3000 copies/mL ( Table 1 ). The presence of severe ultrasound brain abnormalities ( Table 2 ) and the absence of any ultrasound feature of infection or laboratory abnormality in fetal blood were exclusion criteria. The detailed eligibility criteria are listed in the supplementary Appendix . Fetal blood sampling by cordocentesis under ultrasound guidance was offered to all participants to evaluate fetal platelets and viral DNA load to help refine the fetal prognosis. Cordocentesis was not, however, required for study eligibility, as this invasive procedure was not the standard of care in all participating centers.
At least 1 extracerebral abnormality compatible with fetal CMV infection | |
Fetal growth restriction | |
Abnormal amniotic fluid volume | |
Ascites and/or pleural effusion | |
Skin edema | |
Hydrops | |
Placentomegaly >40 mm | |
Hyperechogenic bowel a | |
Hepatomegaly >40 mm | |
Splenomegaly >30 mm | |
Liver calcifications | |
And/or 1 isolated cerebral abnormality | |
Moderate isolated ventriculomegaly (<15 mm) | |
Isolated cerebral calcification | |
Isolated intraventricular adhesion | |
Vasculopathy of lenticulostriate vessels | |
And/or laboratory findings of generalized CMV infection in fetal blood | |
Fetal viremia >3000 copies/mL | |
Fetal platelet count <100,000/mm 3 |
a Since diagnosis of hyperechogenic bowel can be subjective and associated with high interoberver and intraobserver variability, diagnosis of hyperechogenic bowel was only considered for grade-2 or grade-3 hyperechogenic bowel ; this semiquantitative analysis was chosen to limit subjectivity sometimes associated with ultrasound findings.
Presence of at least 1 severe cerebral ultrasound abnormality among following: | |
Ventriculomegaly ≥15 mm | |
Periventricular hyperechogenicity | |
Hydrocephaly | |
Microcephaly <3 SD | |
Mega-cisterna magna >10 mm | |
Vermian hypoplasia | |
Porencephaly | |
Lissencephaly | |
Periventricular cysts | |
Abnormal corpus callosum | |
Or absence of any ultrasound feature of infection or laboratory abnormality in fetal blood |
Procedures
Study design
The trial was a multicenter, open-label, phase II study with 1 arm, based on 1 of Simon’s optimal 2-stage designs. All participants received oral valacyclovir (2 g, 4 times a day, therefore 8 g daily). The medication was continued until delivery or 24 treatment weeks, whichever was sooner. The study drug was purchased from GlaxoSmithKline (Marly Le Roi, France), which had no other role in the study. Visits for clinical and ultrasound examinations, questions about clinical side effects (headache, nausea, neurologic effects listed in the product monograph), and adherence assessment by pill count were scheduled every 2 weeks until delivery. Maternal plasma levels of aspartate aminotransferase, alanine aminotransferase, and creatinine were assessed once a month during treatment. All newborns were examined between days 4 and 7 of life by a trained pediatrician according to a standardized clinical evaluation. Auditory brainstem responses, cranial ultrasound, fundoscopy, and laboratory tests (see criteria for primary endpoint in supplementary data ) were also performed.
The ethics committee of Poissy-Saint Germain Hospital approved the study (2011-001610-34). Participants gave written informed consent before inclusion. Study oversight was provided by an independent data and safety monitoring board (Clinical Research Unit, Cochin-Necker, Paris).
Study endpoints
The primary study endpoint was the proportion of asymptomatic neonates born to women treated with valacyclovir. An asymptomatic neonate was a neonate without growth restriction (that is, with birthweight ≥10th percentile), normal clinical examination, normal laboratory findings, no severe features of infection on cerebral imaging, normal funduscopic examination, and normal audiology findings (see the supplementary Appendix for details).
The secondary endpoints included adverse events related to the study medication and adherence to treatment. CMV DNA levels and platelet counts were compared in pretreatment fetal blood when available and cord blood at birth and in symptomatic and asymptomatic neonates. The effect of the duration of maternal treatment was also assessed.
Historical comparator group
Our systematic review on PubMed using 3 key words (“cytomegalovirus,” “congenital,” “ultrasound”) yielded 216 articles, but only 3 were written in English and included detailed tables that described both prenatal ultrasound findings and neonatal outcome and also included postmortem findings in termination of pregnancy cases. These 3 suitable studies included 724 pregnancies with a maternal CMV primary infection, 217 with an infected fetus. The review of the ultrasound symptoms of these 217 fetuses showed no ultrasound abnormalities in 142 of them and severe cerebral ultrasound abnormalities in 28. Ultrasound abnormalities matching our inclusion criteria were therefore seen in 47 cases, which formed the historical comparator group. Among them 20 (42.55%) neonates were born asymptomatic. Twenty fetuses were terminated and all underwent postmortem examination showing both macroscopic and microscopic evidence of brain damage in all cases. We therefore assumed that these fetuses would have been born symptomatic and classified then into the symptomatic group. See Table S1 for details in the supplementary Appendix .
Laboratory assays
CMV serology and viral load quantification were centralized and analyzed at the Necker Hospital virology laboratory. Maternal CMV primary infection was diagnosed by seroconversion or the concomitant presence of CMV-specific IgM antibodies and low IgG avidity. After extracting DNA from fetal whole blood, cord whole blood, and neonatal urine with the MagnaPure LC platform (Roche Diagnostic, Meylan, France), we performed quantitative polymerase chain reaction for CMV with the CMV-R gene kit (Argene, BioMerieux, Marcy-L’Etoile, France). The limit of detection was 446 copies (or 178 IU)/mL.
Statistical analysis
To estimate the sample size according to Simon optimal 2-stage design, we assumed that a proportion of asymptomatic neonates of <60% was not clinically relevant in relation to valacyclovir efficacy, while a proportion of >80% was deemed acceptable. With a type I error fixed at 0.05 and a power of 80%, we needed 11 infected fetuses for the first stage: if at least 8 cases in stage 1 had a good outcome (were asymptomatic), then 32 additional infected fetuses would be included in stage 2 for a total of 43. If at least 31 of the 43 cases were asymptomatic at birth, valacyclovir would be judged to have a positive effect.
Analyses of efficacy data were performed for all included fetuses (intention to treat). The proportion of asymptomatic neonates and its 95% confidence interval were calculated with an unbiased estimator. In the historical comparator group, we used a random effects model to calculate the overall proportion of asymptomatic neonates and a 95% confidence interval for all 47 cases from the 3 studies.
We used nonparametric tests for all secondary analyses.
Platelet counts at birth and in fetal blood were compared to the gestational age-specific reference range published by Meher-Homji et al. The average individual deviations of values at birth and in fetal blood from the published mean in noninfected fetuses were compared to 0. We also compared individual differences in platelet counts between fetal blood and birth to the slope of the reference mean in noninfected fetuses. Two-sided t tests were used for these analyses.
R software (Version 2.15; R Foundation for Statistical Computing, Vienna, Austria) was used for all statistical analyses and its “meta” package for the metaanalysis.