[1] 42w This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as

[1] 36w In this large cohort study we show that even when prenatal imaging is normal; fetuses infected during the 1 st trimester by CMV are at risk for neurological childhood sequelae including deafness and mild developmental delays.

[1] 97w Congenital cytomegalovirus (CMV) infection is a common prenatal viral infection affecting 0.3-2% of live born infants 1,2 . Fetal infection may follow either maternal primary infection during pregnancy or maternal secondary infection with reactivation of a latent virus or reinfection 3,4 . The diagnosis of primary CMV infection in pregnancy is established after documentation of seroconversion by the new appearance of CMV specific immunoglobulin G (IgG) and immunoglobulin M (IgM) antibodies in the presence of low avidity. Universal screening for CMV infection is the most reliable method to detect primary infection however is considered controversial 5 .

[2] 51w The consequence of intrauterine CMV infection ranges from no adverse short and long term affect to serious congenital malformations, neurodevelopmental delay and sensorineural hearing loss (SNHL). The timing of infection affects the rate of vertical transmission, approximately 30% in first trimester to up-to 70% in the third trimester 6- 9 .

[3] 73w The detection of fetus that will develop severe neurological impairment is of the outmost importance. Serial antenatal ultrasound (US) and 3 rd trimester-magnetic resonance imaging (MRI) are the two modalities used to identify lesions that carry a poor prognosis. A number of studies that compared the efficacy of fetal ultrasound to MRI in predicting fetal outcome have been published [10][11][12][13][14][15][16] , however, all these studies have the disadvantage of a relatively small dataset.

[4] 9w This article is protected by copyright. All rights reserved.

[1] 70w Primary CMV infection during the first trimester with proof of fetal infection on amniocentesis occurred in 123 patients. Of all the participants, 30 had documentation of a maternal symptomatic disease. In one case amniocentesis for proof of fetal infection was not conducted due to the parent's personal preference, however, congenital infection was diagnosed by PCR for the CMV genome in neonatal urine samples during the first few days after delivery.

[2] 199w All patients underwent serial targeted ultrasound examinations and fetal brain MRI was conducted at 32-33 weeks of gestation or earlier if ultrasound examination suspected brain lesions. The number of patients with abnormal fetal ultrasound or MRI findings is illustrated in Table 1. The rate of US findings after first trimester infection was 38/123, 30.9% (12 cases-9.7%, severe brain findings; 15 cases-12.1%, mildmoderate brain findings; 11 cases-8.9%, extracerebral findings including intrauterine growth restriction, hepatomegaly and splenomegaly, echogenic bowel. Classification of findings is based on the Leruez-Ville et al classification 22 ). The rate of abnormal MRI was 37/123 (30.1%); 12/85 (14.1%) for the subgroup of patients with normal US, 21/27 (77.8%) of those with CNS findings on US and 4/11 (36.3%) with non CNS findings on US. The presence of abnormal MRI findings was significantly associated with abnormal sonographic results (p<0.001; +LR 4.47:2.54-7.66). Since most of the MRI findings were hyperintense signals (HIS) in the temporal or frontal lobes (stage 2 and 3 in Table 1), the rates of abnormal MRI after excluding these findings were 21/123 (17.1%), 5/85 (5.9%), 14/27 (51.8%) and 2/11 (18.2%) for all the patients, in the This article is protected by copyright. All rights reserved.

[1] 145w Currently, the consultation of patients with proven fetal 1 st trimester primary CMV infection is based on prenatal findings on ultrasound and MRI. It is clear that severe cerebral injuries demonstrated by these imaging modalities might be associated with a poor neurological outcome, thus requiring a discussion of the possibility of termination of pregnancy in countries where it is legal. However, the prognostic value of non-severe abnormalities on ultrasound and MRI as well as the added value of anomalies identified on MRI imaging alone, with normal serial prenatal ultrasounds, is yet to be determined. As it has been previously suggested that the highest risk of sequelae occurs after first trimester infection 2 ; as demonstrated recently in a large cohort showing no cases of hearing loss or neurologic impairment after second and third trimester 24 , we focused on cases with first trimester primary infection.

[2] 303w The objective of this study was to reassess the contribution of prenatal ultrasound and MR imaging in predicting neonatal and childhood SNHL and neurodevelopmental abnormalities in a fairly large but very strict (only proven 1 st trimester) cohort. In our study we didn't find a correlation between US findings and SNHL or neurodevelopmental abnormalities. For abnormal MRI findings we found a correlation only to neurodevelopmental abnormalities. It should be noted that our data did show a trend with a higher rate of SNHL (more than 2-folds) in the group with abnormal US compared to normal US, however, as mentioned; this was not significant. Larger studies might present a statistically significant correlation. Unlike this study, in previous publications, including our own on a smaller dataset, a correlation was detected This article is protected by copyright. All rights reserved. between abnormal prenatal imaging, both US and MRI, and SNHL and neurological impairment 12,14,22 . One may attribute the fact that we didn't find such a correlation to the relatively high rate of TOPs in the group with abnormal imaging findings (34% of the cases with abnormal US findings and 35% of the cases with abnormal MRI findings). In accordance, when we examined the correlation between composite outcome (including pregnancies that were terminated) and abnormal US and abnormal MRI, we found it to be statistically significant. Of note, the only child in our cohort that developed severe neurological impairment, CP, had severe cerebral abnormalities on prenatal US and stage four cerebral findings on prenatal MRI. In this case the abnormalities on US included underdevelopment of the sulci, dilatation of the ventricles, subependymal cysts, cerebellar malformation, asymmetrical IUGR, cardiomegaly, enlarged stomach and dilated and hyperechogenic bowel. The abnormalities on MRI included dilatation of the ventricles, thinning of the brain parenchyma near the ventricles and choroid plexus cysts.

[3] 236w Currently when prenatal imaging is normal most couples expect a good outcome. In our study out of the 73 cases with both normal US and MRI examinations there were 10 cases of hearing loss (13.7%), including three cases of bilateral complete hearing loss, two of unilateral hearing loss and five cases of partial hearing loss. In addition, there were five cases of neurodevelopmental abnormalities (6.8%, one with concurrent bilateral hearing loss), limited to mild motor delay and hypotonia. Interesting this cases were approximately half of the cases of adverse outcome as altogether we had 18 cases of hearing loss and 12 cases of neurological impairment. Of the studies This article is protected by copyright. All rights reserved. published to date, two focused on the outcome of intrauterine CMV infection with normal prenatal imaging. Amir et al 25 showed a rate of hearing loss of 8.2% at the last assessment and a rate 2% of mild motor delay. Farkas et al 26 focused on normal neurosonography examinations, including 21 patients, and had only one case of bilateral SNHL and no neurological impairment. Both articles included maternal infection throughout the pregnancy (not limited to first trimester) which might have "diluted" the rate of abnormal outcome. Another study, including CMV infection throughout the pregnancy, showed a false negative rate of 5.5% for normal prenatal US and 6.4% for normal prenatal MRI with complications limited to hearing loss 10 .

[4] 130w Compared to these previously published cohorts our dataset showed a higher rate of hearing loss and neurodevelopmental abnormalities when prenatal imaging (both US and MRI) were normal. This might be due to the fact that our cohort was larger or because we focused on first trimester infection which is known to have a higher association to childhood sequelae. Interestingly, in a recently published study of 62 cases of first trimester infection when second trimester assessment (including ultrasound up to 28 weeks and fetal platelet count at cordocentesis) and prenatal MRI were normal the risk for unilateral hearing loss was 16.7% 27 . It should be noted that most neonates with abnormal imaging findings were treated with antiviral therapy that may lower the rate of the development of later childhood sequelae.

[5] 86w The common practice in most centers to date is to offer couples with proven in utero CMV infection both serial US examinations and a 3 rd trimester fetal cerebral MRI. In some centers fetal blood sampling for platelet count and liver enzyme levels are also This article is protected by copyright. All rights reserved. offered 22 . In our study when serial US examinations were normal and MRI identified abnormalities there were only two cases of slight motor delay and one case of partial hearing loss.

[6] 52w The main limitations of our study was the relatively high rate of TOP's, thus preventing us to explore the true statistical association between prenatal findings and clinical sequelae. However, it should be noted, that most TOPs were due to severe abnormal imaging findings which would most likely lead to severe neurological impairment.

[7] 82w The major strengths of our study are its relatively large cohort and its strict inclusion criteria of exact time of onset of maternal infection, routine ultrasound follow-up and the fact that we had MRI examinations in all cases. The data presented in this study should be reviewed during prenatal consultations. However, even though we show that long term sequelae may occur when both prenatal ultrasound and MRI examinations are normal, it should be pointed out that additional larger multicenter studies are necessary.

[8] 90w Figure 1. Outcome according to finding on prenatal imaging. Statistically significant correlation was found between composite outcome (either hearing loss, neurodevelopmental delay or termination of pregnancy) and abnormal prenatal US, abnormal prenatal brain US, abnormal prenatal MRI, abnormal prenatal anatomical MRI and abnormal neonatal brain US and in addition between neurodevelopmental delay and abnormal prenatal MRI. A trend for association was found between hearing loss and abnormal prenatal US and abnormal prenatal brain US and in addition between composite clinical outcome (with hearing loss or neurodevelopmental delay) and abnormal MRI.

[1] 81w In this prospective cohort study, pregnant women with documented primary CMV infection during pregnancy and evidence of fetal infection (vertical transmission) were referred for further evaluation to the Fetal Medicine Unit at Sheba Medical Center, between January 2011 and January 2018. Fifty-six cases included in this study have been reported on in our previous study 14 Dating of pregnancy was precise for all patients and was based on last menstrual period and when necessary corrected according to first trimester US scan.

[2] 80w In Israel, serial screening for CMV infection before or during the first trimester of pregnancy is common. Women who are seronegative in the first trimester are usually reassessed during the second and early third trimesters. Because of this practice, women with serological evidence of primary CMV infection are referred to our center on a routine basis, and data regarding the time of onset of infection and its natural history is available. Few patients were referred for suspected symptomatic maternal infection.

[3] 58w Serological diagnosis of primary CMV infection was documented by IgG seroconversion or presence of IgM antibodies associated with low IgG avidity, as previously described 15,20 . Maternal IgG or IgM assays were performed in various laboratories using enzyme-linked immunosorbent assay. A very low avidity was considered consistent with an onset of infection within the previous 3 weeks. Prenatal

[1] 97w infection are being studied. These include CMV hyperimmune globulin (HIG) 17,18 and the use of in-utero treatment with antiviral therapy 19 . The feasibility of in utero treatment for congenital CMV infection poses a promising solution for this leading cause of neurological impairment. Previous studies have shown that fetal infection in the early stages of pregnancy, mainly first trimester, is associated with higher risk for fetal and neonatal sequelae 8,14 . The aim of this study was to assess the short-and long-term outcome of infected fetuses after maternal first trimester CMV infection in a large prospective cohort.

[2] 9w This article is protected by copyright. All rights reserved.

[3] 104w This article is protected by copyright. All rights reserved. following seroconversion and not earlier than 21 completed week's gestation, as previously described 15 . Patients with proven vertical transmission were counseled again by a fetal medicine specialist regarding risks to the fetus. In one case amniocentesis was not done due to the patients' preference. In this case congenital infection was diagnosed by neonatal urine sample for CMV during the first few days after delivery. We didn't conduct fetal blood sampling for platelet count as described by some groups 19 . In addition, patients didn't receive fetal therapy (including either CMV HIG or antiviral therapy).

[4] 102w Following prenatal detection of fetal CMV infection, targeted ultrasound examination was performed every 3-4 weeks until delivery to assess fetal growth, amniotic fluid volume, placental appearance and ultrasound markers of fetal disease such as ventriculomegaly, echogenic foci or cysts around the ventricles, the posterior fossa and cerebellum, echogenic foci in the liver and echogenic bowel. The ultrasonographic findings were classified according to Ville et al 21 : Severe brain abnormalities, mild to moderate brain abnormalities, and extracerebral findings. High frequency ultrasound equipment was used (GE Voluson 730, E6, E8, E10; GE Medical Systems, Zipf, Austria) for both transabdominal and transvaginal ultrasound examinations.

[5] 140w Fetal brain MRI was performed at 32-33 weeks of gestation or earlier if ultrasound examination suspected brain lesions as previously described 15 up summaries. Follow-up information was obtained at a median age of two (range of six month to 10 years). We didn't have any cases that were lost to follow up. CMV associated sequelae were defined as follows: Bilateral hearing loss (bilateral deafness), unilateral hearing loss (unilateral deafness); partial hearing loss (any clinical reduction in hearing not defined as deafness); neurodevelopmental abnormalities (mainly hypotonia or motor delay); composite clinical outcome includes both any level of hearing loss or neurodevelopment abnormalities; composite outcome includes any level of hearing loss or neurodevelopment abnormalities and in addition pregnancies that were terminated (as it can be assumed that due to the severe findings on imaging these fetuses would most probably have developed sequelae).

[6] 14w The study was approved by the ethics committee of the Chaim Sheba Medical Center.

[7] 49w Statistical analysis was performed using SPSS 16.0 (SPSS Inc., Chicago, IL, USA). Categorical variables were compared using chi-square. Multivariable logistic regression analysis was used to evaluate the independent association of prenatal ultrasound findings (intracranial or other) and MRI findings in relation to the occurrence of any abnormal clinical outcome.

[8] 9w This article is protected by copyright. All rights reserved.

[9] 189w Altogether in our database of 108 live-born children with 1 st trimester infection the rate of hearing loss was 16.6% (18/108; CI 10.6-28.4/108). Twenty neonates (18.5%) had an initial abnormal neonatal hearing test (BERA examination). Of these, three children were later categorized as deaf (all 3/3 with normal prenatal imaging, Table 2), three with unilateral hearing loss (2/3 with normal prenatal imaging) and five with partial hearing loss (3/5 with normal prenatal imaging). Upon follow-up, nine children with initial abnormal BERA had normal hearing. Additional seven children with an initial normal BERA examination developed hearing loss, two with unilateral complete hearing loss and five with partial hearing loss (3/7 with normal prenatal imaging). All, except one child, developed the hearing loss before the age of one year. This child developed partial slight hearing loss around the age of five which was attributed to the prenatal CMV infection. It should be noted that four of the seven children were treated with either Ganciclovir or Valganciclovir, three for abnormal findings on ultrasound examination after birth and one due to the prenatal findings on MRI with no findings on ultrasound after birth.

[10] 103w The rate of neurodevelopmental abnormalities in the cohort was 11.1% (12/108; CI 6.2-20.9/108; 5/12 with normal prenatal imaging). These included mild motor delay and/or hypotonia in 11 cases and slight verbal delay in two cases. There was one case of cerebral palsy (CP). Of the children with neurodevelopmental abnormalities five had in addition hearing impairment, two with bilateral hearing loss (one with normal This article is protected by copyright. All rights reserved. prenatal imaging) and three partial hearing loss. Neurodevelopmental disability was significantly associated with hearing loss (P=0.013). Two infants with attention deficit hyperactivity disorder (ADHD) were not considered to have CMV-associated sequelae.

[11] 158w In the subgroup with abnormal prenatal ultrasound findings the rate of hearing impairment, neurodevelopmental abnormalities and composite outcome was 28%, 20% and 32% respectively. The rate of abnormal prenatal findings on ultrasound was not significantly associated with hearing impairment, neurodevelopmental abnormalities or clinical-composite outcome (p-value 0.084, 0.109 and 0.176 respectively). Since most of the termination of pregnancies were decided upon severe US findings-composite outcome, including TOP's, was significantly associated with abnormal prenatal sonographic findings (p<0.001). The incidence of hearing loss was 7/25 (28%) in the group with abnormal ultrasound findings and 11/83 (13.2%) in the group with normal ultrasound examinations. The incidence of neurodevelopmental disabilities was 5/25 (20%) in the group with abnormal ultrasound findings and 7/83 (8.4%) in the group with normal ultrasound examinations. We didn't find a statistical correlation between abnormal ultrasound finding and childhood sequelae in a subgroup analysis according to the Ville et al classification 23 for severe and mild to moderate brain abnormalities.

[12] 43w In the subgroup with an abnormal prenatal MRI examination the rate of hearing impairment, neurodevelopmental abnormalities and composite outcome was 25%, 25% and 37.5% respectively. We didn't find an association between MRI findings and This article is protected by copyright. All rights reserved.

[13] 145w hearing loss, neither in all cases with an abnormal MRI examination (incidence of 25% for the abnormal MRI group versus 14.3% in the normal MRI group), in the subgroup with anatomical findings (Cannie stage 4 or 5, incidence of 30% versus 14.3%), in the subgroup with HIS (Cannie stage 2 or 3, incidence of 21.4% versus 14.3%) or in the subgroup with a normal ultrasound examination and an abnormal MRI (incidence of 10% versus 13.7%). Interestingly, we did find an association between abnormal MRI examination and neurodevelopmental abnormalities (P=0.014, incidence of 25% for the abnormal MRI group versus 7.1% in the normal MRI group). However, in the subgroup with a normal ultrasound examination and an abnormal MRI we did not find a correlation between abnormal MRI and neurodevelopmental abnormalities (P=0.16, incidence of 20% in the abnormal MRI group versus 6.8% in the normal MRI group).

[14] 51w The rate of SGA at birth in the study was 18/108 (16.6%). Of these only two cases had abnormal extracerebral findings on ultrasounds during pregnancy and three had abnormal MRI examinations (one Cannie stage 4 and two with only HIS). SGA was not associated with an increased risk of CMV-associated sequelae.

[15] 47w Multivariate logistic regression (including prenatal US and MRI findings as well as SGA and neonatal TCUS) did not find any significant predictor for SNHL. Abnormal MRI findings were significantly (p=0.05) associated to neurological abnormalities in the regression model. This article is protected by copyright. All rights reserved.