The investigators' hypothesis is that maternal treatment with Letermovir will inhibit fetal CMV replication better than Valaciclovir in infected fetuses and lead to a higher proportion of negative CMV PCR at birth in neonatal blood collected in the first day of life or in cord blood in case of termination of pregnancy (TOP).

The main objective is to demonstrate that Letermovir administered to women carrying a CMV infected fetus following a maternal infection of the first trimester increases the proportion of neonates with a negative CMV PCR in neonatal blood collected in the first day of life or in cord blood in case of termination of pregnancy (TOP) compared to Valaciclovir.

In each group , the proportion of asymptomatic neonates and the number and type of long-term sequelae at 2 years will also be assessed and compared.

15-20% of CMV infected fetuses are symptomatic and up-to 60% of those symptomatic fetuses have postnatal sequelae. Long-term sequelae are essentially neurological deficiencies and hearing loss. Long-term sequelae are mainly seen in fetuses infected following a maternal infection in the first trimester. The physiopathology of brain and inner ear lesions is not completely elucidated but the viral lesions and viral replication play a major role in this altered neurodevelopment. Fetuses with the most severe brain lesions are also those presenting with high CMV replication in the brain and in all other organs. Moreover, placenta infection affects fetal growth causing growth restriction and therefore affects fetal development in that way. Finally, infected fetuses with high blood viral load at diagnosis (around 22 weeks) are more likely to be symptomatic at birth (OR=5.7 IC95% 2.02-16.53). This correlation between symptoms and high levels of viral replication suggests that an antiviral treatment that could efficiently inhibit viral replication could be beneficial.

Neonatal antiviral treatment with Ganciclovir or Valganciclovir has been used for more than 20 years and is recommended for infected neonates that are symptomatic. Two randomized studies demonstrated that this treatment improves hearing and intellectual capacities of symptomatic neonates with central nervous system involvement. However, this improvement is only modest. This modest benefit can probably be explained by the fact that cerebral lesions developed in utero are already fixed in the neonatal period. The investigators' hypothesis is that early prenatal antiviral therapy for infected fetuses at high risk of cerebral lesions will be more efficient to alleviate long-term sequelae than neonatal treatment. The prognosis of fetal infection can now be established upon fetal imaging by ultrasound (US) and MRI, combined with fetal laboratory tests (fetal platelets count and viral load). The prognosis is poor for severe brain lesions and good when imaging and laboratory parameters are normal. In between these extremes, symptomatic fetuses with extra-cerebral or mild cerebral features are an appropriate target for antiviral therapy with the aim to prevent the development of irreversible cerebral injury.

The 3 antiviral drugs (Ganciclovir, Foscarnet and Cidofovir) that are licensed to treat CMV infection and disease in immunosuppressed patients are nucleotide inhibitors and because of their potential carcinogenicity and teratogenicity, they should be avoided in pregnancy. Valaciclovir is efficient to prevent CMV infection in transplanted patients, is safe in pregnancy and crosses the placenta efficiently. The investigators carried a phase II, not randomized, open label clinical trial to test the efficacy of Valaciclovir in infected fetuses. Valaciclovir was given to women carrying a fetus with at least 1 non-severe ultrasound feature from prenatal diagnosis up until delivery. This led to 79% asymptomatic neonates compared to 43% following natural history of the disease. However, the efficacy of Valaciclovir seemed only partial. First, the antiviral effect was partial: although fetal blood viral load decreased with treatment, 90% of treated fetuses still had detectable CMV DNA in cord blood at birth and all had detectable CMV DNA in neonatal saliva and urine. And second, the clinical efficacy was not optimal since only 57% of fetuses with more than 1 ultrasound feature were born asymptomatic, suggestive of Valaciclovir lower efficacy in such cases.

The investigators therefore looked at new anti CMV drugs. Among them only Letermovir has been licensed to prevent CMV disease in transplanted patients in 2018 and will be available in 2019. Letermovir is not a nucleotide inhibitor and has specific anti-CMV activity. In preclinical toxicity studies it was not genotoxic, not teratogenic and did not impair fertility at the recommended human doses. Besides, no specific concern arises from its safety profile in humans. It controls CMV infection and disease in bone marrow transplant patients by achieving blood viral load clearance in 50-80% of cases.

The investigators' hypothesis is that maternal treatment with Letermovir will inhibit fetal CMV replication better than Valaciclovir in symptomatic infected fetuses and lead to a higher proportion of negative CMV PCR at birth in cord blood. Since severity is largely related to viral replication, clearance of viral replication is a valid surrogate endpoint for clinical outcome in such rare and phenotypically variable cases

The investigators' main objective is to demonstrate that Letermovir administered to women carrying a CMV infected fetus following a maternal infection of the first trimester increases the proportion of neonates with a negative CMV PCR in neonatal blood collected in the first day of life or in cord blood in case of termination of pregnancy (TOP) compared to Valaciclovir. The primary endpoint is the proportion of negative CMV PCR (<500 IU/ml) in neonatal blood collected in the first day of life or in cord blood at termination of pregnancy

The following will also to be compared between the 2 arms : the proportion of asymptomatic neonates, the overall growth, the proportion of long-term sequelae at 2 years of age, the tolerance of treatment for mothers, fetuses and neonates, the maternal adherence to treatment, the evolution of ultrasound features between Day0 and Week 2, Week 4, and Week 6 of treatment, the changes in cerebral and placental features between Day 1st magnetic resonance imaging (MRI) within the first month of inclusion and 2nd MRI at 32 ± 2 WA, the post-mortem examination in cases with medical termination of pregnancy (TOP).

Inclusion Criteria:

• Pregnant woman ≥ 18 years old,
• CMV infection in the 1st trimester
• with an infected fetus at 15 -28 weeks (positive CMV PCR in the amniotic fluid) With a fetus presenting without any severe cerebral ultrasound feature (ventriculomegaly ≥15 mm, hydrocephalus, periventricular hyperechogenicity, microcephaly<-3SD, vermian hypoplasia, porencephaly, lissencephaly, corpus callosum dysgenesis, cystic leukomalacia)
• affiliation to a social security regime//health insurance
• Given consent for the study
• Patient must be able and willing to comply with study visits and procedures

Exclusion Criteria

• Participation to another interventional drug trial (category 1)
• Subject protected by law under guardianship or curatorship
• Maternal CMV infection after 15 weeks'
• Creatinine clearance <50 ml/mn/1,73m²
• Liver insufficiency (Child Pugh grade C), AST, ALT 5 x ULN, bilirubin 2 x ULN.
• Woman with known allergy to Letermovir or Valaciclovir
• Contraindication for the administration of Letermovir and Valaciclovir listed in the SmPC of Prevymis® and Zelitrex®
• Women with hypersensitivity to aciclovir
• Concomitant administration of St John's wort
• Woman treated by pimozide, ergot alkaloids, dabigatran, atorvastatin, simvastatin, rosuvastatin, pitavastatin or cyclosporin.
• Woman with hereditary intolerance to galactose, with lactose lapp deficiency, glucose or galactose malabsorption syndrome