Volume 60, Issue 1 p. 68-75
Original Paper
Free Access

Anatomical and diffusion-weighted imaging of brain abnormalities in third-trimester fetuses with cytomegalovirus infection

M. Aertsen

Corresponding Author

M. Aertsen

Department of Imaging and Pathology, KU Leuven, Leuven, Belgium

Department of Radiology, University Hospitals Leuven, Leuven, Belgium

Correspondence to: Dr M. Aertsen, Department of Radiology, University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium (e-mail: [email protected])Search for more papers by this author
S. Dymarkowski

S. Dymarkowski

Department of Imaging and Pathology, KU Leuven, Leuven, Belgium

Department of Radiology, University Hospitals Leuven, Leuven, Belgium

Search for more papers by this author
W. Vander Mijnsbrugge

W. Vander Mijnsbrugge

Department of Radiology, University Hospitals Leuven, Leuven, Belgium

Search for more papers by this author
L. Cockmartin

L. Cockmartin

Department of Radiology, University Hospitals Leuven, Leuven, Belgium

Search for more papers by this author
P. Demaerel

P. Demaerel

Department of Imaging and Pathology, KU Leuven, Leuven, Belgium

Department of Radiology, University Hospitals Leuven, Leuven, Belgium

P.D. and L.d.C. contributed equally to this study.

Search for more papers by this author
L. De Catte

L. De Catte

Division Woman and Child, Fetal Medicine Unit, Clinical Department of Obstetrics and Gynecology, University Hospital Gasthuisberg, Leuven, Belgium

P.D. and L.d.C. contributed equally to this study.

Search for more papers by this author
First published: 12 January 2022
Citations: 4

Abstract

Objectives

In this study of cytomegalovirus (CMV)-infected fetuses with first-trimester seroconversion, we aimed to evaluate the detection of brain abnormalities using magnetic resonance imaging (MRI) and neurosonography (NSG) in the third trimester, and compare the grading systems of the two modalities. We also evaluated the feasibility of routine use of diffusion-weighted imaging (DWI) fetal MRI and compared the regional apparent diffusion coefficient (ADC) values between CMV-infected fetuses and presumed normal, non-infected fetuses in the third trimester.

Methods

This was a retrospective review of MRI and NSG scans in fetuses with confirmed first-trimester CMV infection performed between September 2015 and August 2019. Brain abnormalities were recorded and graded using fetal MRI and NSG grading systems to compare the two modalities. To investigate feasibility of DWI, a four-point rating scale (poor, suboptimal, good, excellent) was applied to assess the quality of the images. Quantitative assessment was performed by placing a freehand drawn region of interest in the white matter of the frontal, parietal, temporal and occipital lobes and the basal ganglia, pons and cerebellum to calculate ADC values. Regional ADC measurements were obtained similarly in a control group of fetuses with negative maternal CMV serology in the first trimester, normal brain findings on fetal MRI and normal genetic testing.

Results

Fifty-three MRI examinations of 46 fetuses with confirmed first-trimester CMV infection were included. NSG detected 24 of 27 temporal cysts seen on MRI scans, with a sensitivity of 78% and an accuracy of 83%. NSG did not detect abnormal gyration visible on two (4%) MRI scans. Periventricular calcifications were detected on two MRI scans compared with 10 NSG scans. While lenticulostriate vasculopathy was detected on 11 (21%) NSG scans, no fetus demonstrated this finding on MRI. MRI grading correlated significantly with NSG grading of brain abnormalities (P < 0.0001). Eight (15%) of the DWI scans in the CMV cohort were excluded from further analysis because of insufficient quality. The ADC values of CMV-infected fetuses were significantly increased in the frontal (both sides, P < 0.0001), temporal (both sides, P < 0.0001), parietal (left side, P = 0.0378 and right side, P = 0.0014) and occipital (left side, P = 0.0002 and right side, P < 0.0001) lobes and decreased in the pons (P = 0.0085) when compared with non-infected fetuses. The ADC values in the basal ganglia and the cerebellum were not significantly different in CMV-infected fetuses compared with normal controls (all P > 0.05). Temporal and frontal ADC values were higher in CMV-infected fetuses with more severe brain abnormalities compared to fetuses with mild abnormalities.

Conclusions

Ultrasound and MRI are complementary during the third trimester in the assessment of brain abnormalities in CMV-infected fetuses, with a significant correlation between the grading systems of the two modalities. On DWI in the third trimester, the ADC values in several brain regions are abnormal in CMV-infected fetuses compared with normal controls. Furthermore, they seem to correlate in the temporal area and, to a lesser extent, frontal area with the severity of brain abnormalities associated with CMV infection. Larger prospective studies are needed for further investigation of the microscopic nature of diffusion abnormalities and correlation of different imaging findings with postnatal outcome. © 2022 International Society of Ultrasound in Obstetrics and Gynecology.

CONTRIBUTION

What are the novel findings of this work?

Neurosonography (NSG) and magnetic resonance imaging (MRI) are complementary tools in third-trimester assessment of brain abnormalities in cytomegalovirus (CMV)-infected fetuses with seroconversion in the first trimester, and there is a significant correlation between the grading systems of the two modalities. Apparent diffusion coefficient values in different brain regions of CMV-infected fetuses are abnormal when compared with CMV-negative controls. However, there is a significant risk of failure of diffusion-weighted imaging in 15% of routine clinical scans.

What are the clinical implications of this work?

Third-trimester findings on MRI and NSG in fetuses with first-trimester CMV infection can be combined for parental counseling and further management of pregnancy. New classification systems of brain abnormalities in CMV-infected fetuses should incorporate findings detected on diffusion-weighted imaging for a more thorough classification.

Introduction

Cytomegalovirus (CMV) infection is the most common intrauterine infection, with an overall birth prevalence of around 1%1. The overall birth prevalence of symptomatic congenital CMV infection is 0.07%1, 2. Although the transmission rate increases with advancing gestation, the likelihood and severity of fetal impairment is lower when infection occurs later in pregnancy3-5. Signs of fetal infection are classified as extracerebral or cerebral based on imaging. Cerebral signs often occur after extracerebral manifestations6. They include, in descending order of frequency: temporal cysts, other parenchymal abnormalities (calcifications, periventricular abnormalities, T2-hyperintense signal intensity), ventriculomegaly and subependymal cysts and, less commonly, posterior fossa abnormalities, hemorrhage, microcephaly and neuronal migration disorders6-11. These findings are used for counseling and, in cases of severe brain abnormality, may indicate unfavorable prognosis. The presence of non-severe abnormalities or normal findings is more favorable, but may still be associated with long-term neurological sequelae12-14.

Although fetal imaging comprises mainly ultrasonography, distinction should be made between routine screening ultrasound15, 16 and dedicated neurosonography (NSG)17-19. Furthermore, abnormalities can appear ≥ 12 weeks after maternal infection, warranting serial NSG during the remainder of the pregnancy4. Recent studies have shown that even serial NSG does not detect all central nervous system (CNS) abnormalities, particularly when dealing with brain maturation19-22. Magnetic resonance imaging (MRI) is accepted as a complementary tool for the evaluation of the fetal brain7. Its advantages include visualization of the entire brain parenchyma and detection of white matter anomalies18, 20-22. However, evaluation of white matter hyperintense signal on T2-weighted images is subjective and difficult to interpret in the third trimester7, 14. Furthermore, studies evaluating the prognostic value of isolated T2-hyperintense signal of the white matter are contradictory14, 23-25. Diffusion-weighted imaging (DWI) findings and, more specifically, the apparent diffusion coefficient (ADC) have been shown to correlate quantitatively with fetal brain maturation26. To date, little research has been done on the ADC values of the white matter in the CMV-infected fetal brain27, 28.

In this study, we compared the detection of brain anomalies in the third trimester using MRI and NSG in confirmed first-trimester CMV-infected fetuses. The brain anomalies were classified using grading scales for NSG6 and MRI14, and possible correlations between the two modalities were evaluated. Additionally, we investigated the feasibility of routine acquisition of fetal brain DWI during the third trimester in CMV-infected fetuses and compared the ADC values in several anatomical brain areas between CMV-infected fetuses and a control population.

Methods

This was a retrospective review of all MRI scans of fetuses with confirmed CMV infection performed between September 2015 and August 2019 at the University Hospitals Leuven, Leuven, Belgium. Each case had at least one available NSG scan performed in our center by an expert neurosonographer. Confirmed CMV infection was defined as a positive polymerase chain reaction (PCR) analysis of amniotic fluid obtained by amniocentesis at least 8 weeks after estimated maternal infection and after 20 weeks of gestation, according to the recent International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) guidelines4. To analyze the differences in ADC during the third trimester, a control group of CMV-negative fetuses with normal genetic testing was selected from our database. CMV-negative status was defined as negative maternal CMV serology during the first trimester. Detailed information on the CMV-negative cohort can be found in Table S1. This study was approved by the research ethics committee of the University Hospitals Leuven (s64217).

Ultrasound examination

NSG (Voluson E8 or E10 machine (GE Healthcare, Zipf, Austria)) was performed by an expert neurosonographer (L.D.C.) with > 25 years' experience in NSG at a tertiary referral center for fetomaternal medicine, once every 2 weeks after 20 weeks of gestation following diagnosis at our center or once every 2 weeks following referral. This frequency was suggested recently by Leruez-Ville et al.11 and is in agreement with the ISUOG guidelines on congenital infection4. NSG was performed following the ISUOG guidelines17 using convex 2–9-MHz and 4–8-MHz transabdominal transducers. Transvaginal examination was also performed using a 6–12-MHz transvaginal transducer. If the transvaginal approach was not possible (because of breech or transverse position), a linear 3.1–10-MHz transabdominal transducer was used. NSG included two- and three-dimensional imaging in all planes (axial, coronal, sagittal and parasagittal). If fetal position was unfavorable, manual external manipulation was performed to enhance access for ultrasound or rescan after a 30–45 min break. As NSG was performed on a regular basis during pregnancy in these cases, the examination closest to fetal MRI was used for analysis. The neurosonographer was not blinded to previous NSG findings, amniocentesis result or, if already available, MRI findings.

The following variables were collected from the patient's medical report: timing of seroconversion (unknown, first, second or third trimester), availability of amniocentesis, amniotic fluid PCR result (positive or negative for CMV), number of transvaginal scans, ultrasound probes and fetal brain findings on NSG. The NSG report was used to grade brain abnormalities in CMV-infected fetuses on a three-point grading scale6. When no fetal brain abnormality was reported, this was recorded as Grade 0. Mild brain abnormalities (ventriculomegaly < 15 mm, intraventricular adhesions, parenchymal calcifications, subependymal and/or choroid plexus cysts and lenticulostriate vasculopathy) were recorded as Grade 1. Grade 2 was assigned in cases with severe brain abnormality, including ventriculomegaly > 15 mm, periventricular hyperechogenicity, hydrocephalus, microcephaly (defined as head circumference < 2 SD), enlarged cisterna magna, vermian hypoplasia, porencephaly, lissencephaly, periventricular white matter cystic lesions and agenesis of the corpus callosum.

MRI examination

All MRI acquisitions were performed using a 1.5T MR imaging system (Siemens Aera, Erlangen, Germany) with two small body coils placed adjacent to each other over the maternal abdomen. The mother was positioned in the supine or left lateral position. No maternal sedation was used. The routine protocol includes T2-weighted fast-spin echo sequence (repetition time (TR), 1000 ms; echo time (TE), 133 ms; field of view (FOV), 300–380 mm; slice thickness (ST), 3 mm; no gap) in the sagittal, axial and coronal planes relative to the fetal head, axial T1-weighted turbo-spin echo sequence (TR, 3000 ms; TE, 4.93 ms, FOV, 350 mm; ST, 4 mm; 1 mm gap) and echo planar imaging (EPI) sequence (TR, 4810 ms; TE, 51.0 ms, FOV, 350 mm; ST, 4 mm; 0.4 mm gap). The latter was used for the detection of susceptibility artifacts to pick up bony structures, blood breakdown products or iron accumulation29. DWI of the fetal head was acquired in the axial plane with a b-value of 1000 s/mm2 (TR, 4900 ms; TE, 63.0 ms, FOV, 240 mm, ST, 4 mm; 1 mm gap) in all cases included in this study. All fetal MRI examinations were supervised by a radiologist experienced in fetal MRI (M.A.), approving image quality for all anatomical sequences, i.e. requesting repetition if necessary due to fetal or maternal motion, as recommended by the ISUOG guidelines30. Timing of MRI was based on NSG findings: if NSG was normal or demonstrated mild abnormalities, MRI was performed at 28–30 weeks of gestation; if severe abnormalities were noted on NSG, MRI was performed earlier to provide the best parental counseling as early as possible, taking into account the local legislation regarding termination of pregnancy. The radiologist had full access to the patients' medical file containing information regarding previous NSG, amniocentesis and, if available, MRI report.

The following variables were analyzed: temporal lobe abnormalities, classified as temporal horn dilatation, increased temporal signal intensity or temporal cyst according to Doneda et al.8, presence of occipital cysts, white matter destruction, germinal matrix hemorrhage, cortical abnormalities and other abnormalities (e.g. calcifications, lenticulostriate vasculopathy)7. All fetal MRI images were reported by the same radiologist (M.A.) with > 5 years' experience in fetal MRI in a tertiary referral center for fetomaternal medicine. Typical brain abnormalities suggestive of CMV infection were recorded8, 31. MRI abnormalities were graded according to Cannie et al.14: Grade 1, no abnormality; Grade 2, isolated periventricular T2-weighted signal hyperintensity in the frontal and parieto-occipital areas; Grade 3, isolated T2-weighted hyperintense signal in the temporal lobes; Grade 4, cysts and/or septa in the temporal and/or occipital lobes; and Grade 5, migration disorders, cerebellar hypoplasia, microcephaly. Signal hyperintensity was determined by subjective evaluation, taking into account gestational age and normal maturational processes. Hypoplasia and microcephaly were defined as biometric parameters below the 5th percentile on reference charts32, 33. Cortical abnormality was defined as cortical lining abnormality and loss of symmetry with or without abnormal periventricular brain parenchyma on T2-weighted imaging present in at least two planes34-38.

Quality of DWI sequence was scored subjectively by the same observer (W.V.M.) as poor (different areas cannot be identified), suboptimal (areas can be delineated, but moderate noise present), good (areas can be seen, but some noise present) or excellent (all areas are sharp and can be identified easily). A freehand region of interest (ROI) was placed in the white matter areas of the frontal, parietal, temporal and occipital lobes and the basal ganglia, pons and cerebellum on the ADC map provided by the system to obtain ADC values (Figure 1). Similar regional ADC measurements were obtained from a control group of CMV-negative fetuses without brain abnormalities on imaging. All ADC measurements in the CMV and control cohorts were performed by the same observer (W.V.M.) and checked by a radiologist experienced in fetal MRI (M.A.).

Details are in the caption following the image
Axial apparent diffusion coefficient images in a fetus at 34 weeks, showing regions of interest drawn freehand at the cerebellar hemispheres (a), pons (b), temporal white matter (c), occipital white matter (d), basal ganglia (e), frontal white matter (f) and parietal white matter (g).

Statistical analysis

Descriptive data are given as mean ± SD or n (%), depending on the parameter. For the detection of different abnormalities on NSG and MRI, sensitivity and specificity were calculated based on the gold standard in clinical routine. MRI was considered to be the gold standard for the detection of temporal lesions, as it is able to differentiate between temporal horn dilatation, increased parenchymal signal and cysts8. To compare the performance of the two modalities in temporal cyst detection, only MRI abnormalities classified as temporal cyst according to Doneda et al.8 were considered in the comparison with NSG findings. Gyration abnormalities are detected more easily on MRI, especially in the third trimester18. Although MRI uses EPI and T1-weighted imaging to increase the detection of calcifications, ultrasound is more accurate in the detection of calcifications and lenticulostriate vasculopathy39. Correlation between NSG vs MRI grading was evaluated using a non-parametric Spearman rank correlation test.

Comparison of the ADC values between CMV-infected fetuses and the controls was performed using the Mann–Whitney U-test. Additionally, the Mann–Whitney U-test was used to investigate differences in ADC values between fetuses with different characteristics within the CMV cohort. Statistical analysis was performed using Analyze-it® (Analyze-it for Microsoft Excel 4.81.4; Analyze-it Software Ltd, Leeds, UK). A P-value < 0.05 indicated statistical significance.

Results

Fifty-three MRI examinations of 46 CMV-infected fetuses with confirmed first-trimester CMV infection were included (Figure S1, Table 1). The mean body mass index of the group was 25.74 ± 3.0 kg/m2. The mean gestational age was 30.1 ± 2.4 weeks at MRI and 29.6 ± 2.3 weeks at NSG. On average, NSG was performed 0.5 ± 0.7 weeks before MRI. In four cases, the included NSG scan was performed after MRI, but all of them underwent several previous NSG examinations before MRI. All fetuses with two MRI examinations underwent transvaginal ultrasound. In total, 38 (83%) fetuses were evaluated on transvaginal ultrasound. Eight (17%) cases that could not be evaluated by the transvaginal approach were evaluated using the transabdominal linear 3.1–10-MHz transducer in addition to the standard transabdominal evaluation using a convex transducer.

Table 1. Brain pathology in 46 cytomegalovirus-infected fetuses detected on magnetic resonance imaging (MRI) and neurosonography (NSG) in the third trimester
Pathology MRI NSG
Temporal lesion 40 (75) 29 (55)
Temporal horn dilatation* 1/40 (3)
Increased temporal signal intensity* 12/40 (30)
Temporal cyst* 27/40 (68) 24/29 (83)
Calcifications 2 (4) 10 (19)
Lenticulostriate vasculopathy 0 (0) 11 (21)
White matter destructive cysts 18 (34) 12 (23)
White matter hyperintensity on MRI/periventricular halo on NSG 36 (68) 12 (23)
Germinal matrix hemorrhage 0 (0) 0 (0)
Cortical abnormality 2 (4) 0 (0)
  • Data are given as n (%) or n/N (%).
  • Percentages are calculated based on number of NSG and MRI scans (both n = 53).
  • * Type of temporal lesion on MRI, according to Doneda et al.8.

NSG vs MRI

Temporal cyst was detected on 24 NSG and 27 MRI scans. In three cases, cysts were detected on NSG but not on MRI performed 1 week later; these were considered false positives. With 21 true-positive, six false-negative and 23 true-negative scans, NSG had a sensitivity of 78% for the detection of temporal cysts detected on MRI, with a positive predictive value of 88% and an accuracy of 83%.

NSG did not detect abnormal gyration visible on two (4%) MRI scans. Periventricular calcifications were detected on two (4%) MRI scans compared to 10 (19%) NSG scans. While lenticulostriate vasculopathy was detected on 11 (21%) NSG scans, no fetus demonstrated this finding on MRI. An overview of the abnormalities detected using each modality can be found in Table 1.

When comparing the NSG6 and MRI14 grading of brain abnormalities, there was a significant correlation between the two grading systems (rs = 0.788; P < 0.0001). An overview of the number of cases included in each group per modality is given in Table 2. Four scans were graded as normal on both MRI and NSG.

Table 2. Severity classification of third-trimester findings on fetal neurosonography (NSG) and magnetic resonance imaging (MRI) in fetuses with confirmed cytomegalovirus infection
MRI NSG
Grade 1 5 (9) Grade 0 21 (40)
Grade 2 9 (17) Grade 1 15 (28)
Grade 3 12 (23)
Grade 4 25 (47) Grade 2 17 (32)
Grade 5 2 (4)
  • Data are given as n (%).
  • Percentages are calculated based on number of NSG and MRI scans (both n = 53).
  • NSG findings were graded according to Leruez-Ville et al.6.
  • MRI findings were graded according to Cannie et al.14.

Apparent diffusion coefficient

Thirty-four CMV-negative fetuses with 36 MRI examinations between 23 and 38 weeks of gestation (mean ± SD gestational age at MRI, 29.6 ± 3.5 weeks) were included in the control group. There was no statistical difference in gestational age at MRI between the control and CMV-infected groups (P = 0.07). The mean maternal age was 31.5 ± 4.7 years in the control group and 30.7 ± 4.1 years in the CMV-positive group. The clinical indication for fetal MRI is provided in Table S1.

In the CMV-infected group, six (11%) of the DWI scans were of poor quality and two (4%) were of suboptimal quality and were therefore excluded (Figure 2). All scans classified as good (7/53; 13%) or excellent (38/53; 72%) were used for further analysis.

Details are in the caption following the image
Axial apparent diffusion coefficient maps of the brain in four fetuses, demonstrating the variable quality of images. (a) Poor-quality image with a lot of noise, artifacts and loss of anatomical differentiation. (b) Suboptimal-quality image with a lot of noise but no artifacts and with some differentiation of different anatomical areas. Good-quality (c) and excellent-quality (d) images with minimal and no noise, respectively, and with much better distinction of the different anatomical areas. Only scans scored as good or excellent were included for further analysis.

Pathology detected on ultrasound and potentially included in the ADC ROIs was lenticulostriate vasculopathy in the ROI placed at the basal ganglia (n = 11) and periventricular calcifications in the ROI of the parietal or frontal white matter (n = 10). ROIs in the temporal, occipital and frontal lobes were placed in such a way that periventricular cysts were avoided. This was not possible for calcifications because of the very low sensitivity of MRI. The periventricular hyperechogenicity (n = 12) was included in the parietal white matter ROI.

ADC values of CMV-infected fetuses were increased significantly in the frontal, temporal, parietal and occipital lobes and decreased in the pons when compared with CMV-negative fetuses. In the basal ganglia and cerebellum, the ADC was not significantly different between CMV-infected fetuses and controls. Mean ADC values for each cohort are provided in Table 3.

Table 3. Apparent diffusion coefficient values in different brain regions in fetuses with confirmed first-trimester cytomegalovirus (CMV) infection and CMV-negative controls
Region Controls (n = 36*) CMV infected (n = 45*) P
Frontal white matter L 1618.90 ± 88.30 1752.20 ± 56.10 < 0.0001
Frontal white matter R 1618.17 ± 89.89 1766.98 ± 141.58 < 0.0001
Parietal white matter L 1659.00 ± 101.99 1722.58 ± 160.67 0.0378
Parietal white matter R 1651.14 ± 90.43 1735.20 ± 126.44 0.0014
Occipital white matter L 1551.03 ± 103.36 1676.13 ± 148.21 0.0002
Occipital white matter R 1549.08 ± 113.01 1684.93 ± 147.83 < 0.0001
Temporal white matter L 1545.03 ± 84.26 1654.18 ± 111.05 < 0.0001
Temporal white matter R 1555.00 ± 84.40 1668.22 ± 113.12 < 0.0001
Basal ganglia L 1233.11 ± 79.87 1233.27 ± 78.30 0.9394
Basal ganglia R 1236.19 ± 81.86 1233.00 ± 77.61 0.7073
Cerebellum L 1305.67 ± 110.00 1345.40 ± 108.02 0.1929
Cerebellum R 1302.11 ± 100.12 1349.87 ± 112.02 0.0962
Pons 1187.94 ± 11.12 1126.13 ± 60.72 0.0085
  • Data are given as mean ± SD.
  • * Number of scans. L, left; R, right.

In the four scans with good/excellent quality DWI imaging and normal findings on NSG and MRI, all ADC measurements in regions with a significant difference in ADC values between the two cohorts were between the 5th and 95th percentiles. Among the 41 scans with anatomical abnormalities on MRI and/or NSG and with good/excellent quality DWI imaging, only 15 (38%) had a normal ADC value in the frontal white matter region, while ADC in other brain regions was normal in ≥ 50% of scans (Table 4).

Table 4. Proportion of scans with apparent diffusion coefficient (ADC) values in the normal range (between the 5th and 95th percentiles) among scans with abnormalities on neurosonography and/or magnetic resonance imaging (n = 41)
Region Normal ADC range (5th–95th percentile) Normal ADC (n (%))
Frontal white matter L 1474–1748 15 (37)
Frontal white matter R 1491–1752 15 (37)
Parietal white matter L 1516–1846 22 (54)
Parietal white matter R 1500–1816 27 (66)
Occipital white matter L 1340–1692 23 (56)
Occipital white matter R 1377–1713 26 (63)
Temporal white matter L 1401–1697 25 (61)
Temporal white matter R 1411–1660 20 (49)
Pons 1018–1374 38 (93)
  • L, left; R, right.

When comparing the ADC values in the temporal regions between CMV-infected fetuses with (32 scans) and those without (13 scans) temporal lesions, the ADC appeared to be higher in the former group, though the difference was not significant. In CMV-infected fetuses without compared to those with temporal lesions, the mean ± SD ADC values were, respectively, 1627.62 ± 126.15 vs 1684.72 ± 105 in the right temporal lobe (P = 0.115) and 1610.69 ± 106.78 vs 1671.8 ± 109.44 in the left temporal lobe (P = 0.133). The ADC values in the left frontal lobe were significantly higher in CMV-infected fetuses with Grade-4 or -5 abnormalities on T2 imaging (1814.30 ± 167.37) compared to CMV-infected fetuses with Grade 1–3 (1705.86 ± 12.64; P = 0.013); however, there was no significant difference in the ADC values in the right frontal lobe between the two groups (1805.87 ± 135.89 vs 1726.32 ± 138.8; P = 0.069). The right and left temporal ADC values were significantly higher in fetuses with Grade-4 or -5 abnormalities compared to fetuses with Grade-1–3 lesions (right side, 1692.96 ± 116.38 vs 1613.64 ± 91 (P = 0.023); left side, 1705.96 ± 110.86 vs 1628.77 ± 103.66 (P = 0.026)). This was not seen in the other brain regions (all P > 0.05).

Discussion

NSG is the most important modality for detection of fetal brain abnormalities40, 41. Our patients underwent several NSG scans prior to MRI examination around 30 weeks of gestation, in keeping with other studies and to maximize its contribution with respect to local legislation regarding termination of pregnancy18, 19, 41. We demonstrated the added value of MRI in the third trimester for detection of brain abnormalities, which was able to detect more cases with temporal lesions and neuronal migration abnormalities. The latter were unilateral cortical irregularities in the parietal lobe with loss of symmetry and subtle T2-hypointense signal in the underlying white matter34. On the other hand, NSG has proven to be superior in the detection of lenticulostriate vasculopathy and calcifications despite the use of EPI-sequences to enhance MRI detection. In addition, we have found a correlation between NSG6 and MRI14 grading systems for brain abnormalities, whereby severe lesions on NSG were also graded as severe on MRI. The more extensive grading scale of MRI compared with that of NSG as well as the ability of MRI to detect temporal signal abnormalities14 is reflected by the scoring of mild abnormalities, similar to Lipitz et al.13.

DWI is applied regularly for the investigation of CNS abnormalities but remains a challenging technique27, 42-53. In our CMV cohort, we had a successful DWI acquisition in 85% of examinations, justifying routine implementation, but the possibility of failure should be discussed. We believe that the success rate would be > 95% with a second attempt.

Compared with controls, our CMV cohort had significantly different ADC values in the pons and all white matter regions. Furthermore, in agreement with Yaniv et al.28, there was a significant difference in the ADC values in the temporal lobe between fetuses with mild and those with severe CMV brain abnormalities. The frontal ADC was significantly different between groups only on one side.

Kotovich et al.27 found lower ADC values in CMV-positive fetuses with unremarkable fetal MRI when compared with controls, which was in contrast to our findings, as we detected normal ADC values in CMV-positive fetuses with normal NSG and MRI findings. In addition, at least one-third of fetuses with brain abnormalities had ADC values within the normal range in several regions. This variability could be of interest for a more thorough stratification by severity. In our experience, DWI can help in the counseling of parents with a fetus with isolated white matter hyperintensity, as this is known to be associated with a good postnatal outcome, provided that it is not present in the temporal lobe14. Furthermore, it might help in the analysis of fetuses with temporal lesions, as only half of them are affected by sensorineural hearing loss and 25% by neurological impairment14. Nevertheless, the association between DWI abnormalities and clinical manifestations in fetuses with first-trimester CMV infection remains to be studied.

The finding of the increased white matter ADC values in CMV-infected fetuses is in contrast to the decreased ADC values reported by previous studies27, 28, 50. This can be attributed to the difference in study cohorts, as this study focused on the confirmed first-trimester CMV-infected fetuses, while others included fetuses with CMV infection in all trimesters and even those with unknown time of infection27, 28, 50. In addition, we scanned earlier in gestation compared with Yaniv et al.28. These differences in timing are likely to have influenced the presence of acute and chronic abnormalities. Acute inflammation is evident as edema and inflammation (ADC decrease), while chronic infection results in gliosis (ADC increase); both have been detected in CMV-infected fetal brains54. We confirmed this time dependency in the limited number of postmortem examinations in our series. For example, in one case, the histological evaluation revealed acute inflammation. The ADC values in the frontal and parietal white matter of this fetus were decreased. Another case demonstrated necrotic areas as well as inflammatory cells surrounded by gliosis. In this case, the ADC values of the white matter in the frontal region were elevated and those in the parietal region were reduced. Thus, ADC was able to quantify these white matter differences, which are likely to influence neurodevelopmental outcome. All our findings are in line with a previous study comparing white matter abnormalities between CMV-infected neonates and neonates with periventricular leukomalacia55. The study demonstrated similar changes in MR parameters in both cohorts compared with controls, suggesting that the increased ADC is representative of white matter lesions due to axonal loss, abnormal myelin deposition and astrogliosis. However, the location of white matter abnormalities differed between the two cohorts31, 56, 57.

We acknowledge several limitations of this study. First, we included a smaller homogeneous cohort of fetuses with confirmed first-trimester CMV infection over a larger heterogeneous cohort because all long-term sequelae develop after a first-trimester fetal infection5, 58. Second, the retrospective design precluded standardized postnatal or postmortem follow-up. In addition, it might have caused bias because some fetuses with severe brain abnormalities on NSG in early pregnancy did not undergo MRI before termination of pregnancy. Third, the radiologist and neurosonographer were not blinded to previous results. As MRI is usually performed after NSG, this favors the detection of abnormalities by the radiologists; however, this reflects routine clinical care. Last, our control population had normal brain development at the time of imaging, negative first-trimester maternal CMV serology and normal genetic testing, but postnatal developmental outcome was unknown in the majority of cases.

Future research should aim to improve the assessment of fetuses at risk of symptomatic CMV infection. NSG remains the main modality for lesion detection, but this study demonstrates that MRI with DWI may be of value in the third-trimester assessment. This needs to be investigated in a larger prospective study evaluating the association between neurological sequelae and prenatal imaging findings on NSG and MRI with DWI. In view of therapeutic advances, a standardized maternal serological screening program should help define the population that would benefit from therapy that limits vertical transmission59, 60.

In conclusion, NSG and MRI are complementary in the third-trimester detection of brain abnormalities in CMV-infected fetuses, with a significant correlation between grading systems of the two modalities. On DWI, the ADC values in different brain regions of CMV-infected fetuses are abnormal compared with CMV-negative controls. Furthermore, they seem to correlate in the temporal and, to a lesser extent, frontal areas with the severity of the brain abnormalities.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.