Congenital cytomegalovirus infection: patterns of fetal brain damage

Authors


Corresponding author: L. Gabrielli, Operative Unit of Clinical Microbiology, St Orsola-Malpighi University Hospital, Via Massarenti 9, 40138 Bologna, Italy
E-mail: liliana.gabrielli@aosp.bo.it

Abstract

Clin Microbiol Infect 2012; 18: E419–E427

Abstract

Cytomegalovirus (CMV) is the most prevalent infectious agent causing neurological dysfunction in the developing brain. This study analysed the different patterns of tissue damage, particularly in the brain, of fetuses with documented CMV infection. We studied 45 fetuses at 20–21 weeks of gestation with congenital CMV infection documented by invasive positive prenatal diagnosis. At the time of amniocentesis, abnormal ultrasound findings had been recorded for 13 of the 45 fetuses (29%). Histological and immunohistochemical characterization was performed on the placenta, brain, heart, lung, liver, kidney, and pancreas. The different degrees of brain damage were correlated with tissue viral load, inflammatory response, placental functionality, and extramedullary haematopoiesis. Even though a high CMV load was detected in all amniotic fluids, brain infection occurred in only 62% of the fetuses and with different degrees of severity. Tissues with a low viral load showed a globally weak inflammatory response, and fetuses had only mild brain damage, whereas tissues with a high CMV load showed prominent infiltration of the activated cytotoxic CD8+ T-lymphocytes responsible for immune-mediated damage. Furthermore, severe placental infection was associated with diffuse villitis and necrosis, consistent with functional impairment and possible consequent hypoxic cerebral damage. Brain injury induced by CMV congenital infection may be the result of uncontrolled viral replication, immune-mediated damage by cytotoxic CD8+ T-lymphocytes, and, in the presence of placental insufficiency, fetal hypoxia.

Introduction

Congenital cytomegalovirus (CMV) infection is the leading cause of congenital viral infection and brain disease in children in developed countries, occurring in approximately 1% of all live births (from about 0.3% to 2.4%). Mother–child transmission is mainly the result of primary maternal CMV infection, which carries a risk of transmission varying from 14.2% to 52.4% (combined prevalence of 32.4%) [1]. Ten per cent of congenitally infected newborns are symptomatic at birth, and permanent neurological injury occurs in up to 60% of these infants [2,3]. Even with antiviral therapy, these injuries are often irreversible. Neurological outcomes may include cerebral palsy, mental retardation, sensorineural hearing loss, seizures, and visual impairments. Microcephaly, periventricular calcifications, hydrocephalus, cerebellar atrophy and polymicrogyria have been shown to be strong predictors of an adverse neurological outcome [4].

The pathogenesis of injury to the developing fetal central nervous system (CNS) is unknown. Multiple mechanisms have been proposed, namely: CMV acting as a ‘teratogen’, the impact on apoptosis, the critical developmental windows of susceptibility, the role of the immune response and the inflammatory processes in potentiating CNS injury, and the potential pathogenic impact of CMV on the endovascular system [5].

A recent study by our group [6] found that fetal brain infection occurred in 55% of cases and with different histological features when a high viral load was present in the amniotic fluid. Correlations between inflammatory infiltrate and tissue damage and between placental and brain damage were observed.

The purpose of the present study was to provide more details on the marked difference in the patterns of tissue brain damage in fetuses with congenital CMV infection documented at 20–21 weeks of gestation from primarily infected mothers. We characterized the fetal inflammatory response and investigated tissue viral load to search for a relationship between the severity of tissue infection and histological lesions. Placental functionality and extramedullary haematopoiesis were studied to evaluate the possible role of hypoxic injury in brain damage.

Materials and Methods

Patients

Forty-five fetuses with congenital CMV infection documented at 20–21 weeks of gestation by invasive positive prenatal diagnosis underwent histological examination.

All of the fetuses were from pregnant women with primary CMV infection arising before the 12th week of gestation. Women who had anti-CMV IgM and anti-CMV IgG of low avidity or who seroconverted to CMV IgG positivity were classified as having primary infection [7–9].

The diagnosis of fetal CMV infection was based on CMV positivity in amniotic fluid by culture and by real-time PCR at 20–21 weeks of gestation. The viral load was >105 copies/mL in all amniotic fluids.

At the time of amniocentesis, all pregnant women underwent ultrasound examinations that included a survey of all fetal organs [10].

Given the particular neurotropism of CMV, the intracranial anatomy was evaluated by targeted transvaginal neurosonographic examination, when permitted by the low position of the fetal head [11].

As negative controls, we studied ten fetuses of the same gestational period from CMV-seronegative women. One was a spontaneous miscarriage for cervical incompetence, and the others were elective terminations of pregnancy because of cardiac malformations (three cases), spina bifida (two cases), trisomy 21 (three cases), and trisomy 18 (one case).

The fetal tissues were analysed after informed consent had been obtained from the parents and according to the policies of the Ethical Committee of St Orsola-Malpighi University Hospital, Bologna, Italy and the rules of the Italian Ministry of Health, Rome, Italy.

Histological examinations

Fetal tissues were fixed in buffered 4% formaldehyde, and haematoxylin and eosin standard sections were obtained from paraffin-embedded blocks of the placenta, pancreas, kidney, lung, liver, brain, and heart.

Immunohistochemical staining for CMV early (ppUL44) antigen (EA) was performed to identify CMV-positive cells. The inflammatory infiltrate (B-lymphocytes and T-lymphocytes) was studied by immunohistochemical staining for CD3, CD4, CD8, and CD20. Granzyme B expression was investigated as a marker of cytotoxic activation of immune effector cells. Cleaved caspase-3 was studied to evaluate apoptosis. Extramedullary haematopoiesis was identified in the liver, pancreas and kidney, with immunohistochemical staining for terminal deoxynucleotidyl transferase (TDT), glycophorin C, and myeloperoxidase.

The following antibodies were used: CMV EA (clones CCH2 and DDG9), anti-CD8 (clone cd8/144b), anti-CD20 (clone L 26), anti-glycophorin C (clone Ret 40f), anti-myeloperoxidase (polyclonal) and anti-TDT (polyclonal) from Dako (Glostrup, Denmark); anti-CD4 (clone IF6) from Novocastra (Newcastle, UK); anti-granzyme B (clone GrB7) from Monosan (Leiden, The Netherlands); anti-CD3 (rabbit monoclonal) from Thermo Scientific (Fremont, CA, USA); and anti-cleaved caspase-3 (rabbit monoclonal) from EuroClone (Milan, Italy).

Histologically, neuropathological lesions were classified according to a semiquantitative score that considered the extent and severity of injury, as in other studies on viral encephalitis [12,13].

Cerebral damage in CMV-positive brains was assessed by combining the tissue lesions with distribution, considering the following separate pathological findings for each brain region (cortex, white matter, and germinal matrix): tissue necrosis (0 = absent; 1 = focal; 2 = diffuse), microglial nodules (0 = absent; 1 = occasional; 2 = scattered; 3 = multiple). A composite score was derived by summing the individual scores of the three regions (ranging from 0 to 15).

CMV-negative areas from the cortex to the germinal matrix in CMV-positive brains were also analysed to find more subtle evidence of tissue damage, such as: microglial activation (0 = absent; 1 = focal; 2 = diffuse) and astrocytosis (0 = absent; 1 = focal with ramified astrocytes; 2 = diffuse with ramified astrocytes; 3 = 1 or 2 plus astrocyte nuclear clearing—Azheimer’s type II astrocytes—and/or apoptosis) and vascular changes (0 = no changes; 1 = focal increased number of vessels; 2 = diffuse increased number of vessels, 3 = 1 or 2 plus plump endothelial cells). The final score was the sum of the individual scores (ranging from 0 to 24).

As a whole, brain damage was evaluated by summing the two different cerebral injuries outlined above on a scale ranging from 0 to 39. This scale identified three groups of fetuses: group A, fetuses with severe cerebral damage (score 25–39); group B, fetuses with moderate cerebral damage (score 16–24); and group C, fetuses with minor cerebral damage (score 0–15).

The inflammatory infiltrate was scored semiquantitatively to assess the percentage of CD8 T-lymphocytes as compared with the surrounding cells. Five fields at ×20 were observed, for each brain region analysed, and the percentage score as the follows: score 0, no lymphocytes detected; score 1, 1–10%; score 2, 11–25%; score 3, 26–50%; and score 4, >50%.

To each of these scores, we added another one according to the distribution of lymphocytes, such as: score 1, scattered lymphocytes; score 2, scattered lymphocytes and clusters; score 3, scattered lymphocytes, clusters, and rare microglial nodules; and score 4, scattered lymphocytes, clusters, and multiple microglial nodules.

As a marker of activated CD8 T-lymphocytes, granzyme B was graded as a percentage of the lymphocytes observed: score 0, no activated CD8 T-lymphocytes; score 1, 1–10%; score 2, 11–25%; score 3, 25–50%; and score 4, >50%.

As a whole, the inflammatory response was evaluated by adding the three scores in each field examined and dividing by five to obtain a median value: the final score ranged from 0 to 12. A mild inflammatory response was between 0 and 3, a moderate response was between 4 and 8, and a severe response was between 9 and 12.

All assessments were performed blindly and independently by two authors (M.P.B. and D.S.), and this was followed by simultaneous viewing and resolution of any discrepancies.

Determination of placental function

To evaluate the extent of placental tissue function, we examined ten histological fields at ×10 magnification in each case in haematoxylin and eosin. We considered villi showing hydrops, necrosis and inflammatory infiltrate to be non-functional.

Virological examinations

The shell-vial procedure was used for CMV isolation from amniotic fluid as described elsewhere [8]. DNA was extracted from 1 mL of amniotic fluid with the NucliSens easyMAG System (bioMerieux, Marcy l’Etoile, France). DNA extraction from paraffin-embedded tissue was performed on 5-μm tissue slices with the BioSprint 15 DNA Blood Kit (Qiagen, Hilden, Germany). Purified DNA was eluted in 50 μL of distilled water, and quantified with real-time PCR assays (Quantifiler Human DNA Quantification Kit; Applied Biosystems, Foster City, CA, USA). Five nanograms of DNA was processed for CMV DNA quantification.

CMV DNA was quantified with a real-time PCR assay (Q-CMV Real Time Complete Kit; Nanogen Advanced Diagnostics SRL, Turin, Italy). Amplification, detection and analysis were performed with the ABI PRISM 7300 platform (Applied Biosystems). The detection limit for amniotic fluid was 1.1 copies/10 μL, and viral load was reported as number of copies/mL. The detection limit for tissue samples was 10 copies/5 ng of DNA, and viral load was reported as number of copies/5 ng of DNA.

Statistical analysis

As a non-parametric significance test, the Mann–Whitney test was performed.

Results

In the 45 fetuses studied, CMV-positive cells were detected in all placentas, 90% of pancreases, 82% of lungs, 80% of kidneys, 67% of livers, 62% of brains, and 34% of hearts. As expected, no CMV EA-positive cells were observed in the tissues of the ten control fetuses from CMV-seronegative women.

We found no evidence of brain CMV infection by immunohistochemistry in 17 cases (38%), whereas 28 cases (62%) were CMV EA-positive. No histological damage was observed in either the control cases or the 17 fetuses lacking evidence of brain infection.

In contrast, the 28 fetuses with CMV-positive brains were classified into three groups, as described in Materials and Methods. The following groups were identified: group A, seven fetuses with severe cerebral damage; group B, 15 fetuses with moderate damage; and group C, six fetuses with mild damage.

Placental function, immune response, extramedullary haematopoiesis and tissue viral load were studied in the fetuses of each group.

Group A: seven fetuses with severe cerebral damage (score 25–39)

Histological findings.  Brain: CMV-positive cells were found in the cortex, white matter, germinal matrix, grey matter, ependyma, and leptomeninges. The brainstem and cerebellum were less often involved. CMV-positive cells comprised neurons, neuroblasts, glia, endothelial, ependymal and meningeal cells.

The cerebral cortex showed extensive laminar necrosis in the third layer, where the developing neurons were replaced by abundant macrophages (Fig. 1). In some areas, the cortex had the appearance of early polymicrogyria, with deepening of the immature sulci. White matter showed oedema, periventricular leukomalacia, increased vascularity with plump endothelium, glial karyorrhectic bodies, abundant macrophages, and areas of ‘ferrugination’ (deposition of iron and calcium in neurons, axons, and dendrites). The ependymal lining was also affected, with multiple areas of ventriculitis. The inflammatory infiltrate was severe (score 9–12) and mainly composed of multifocal aggregates of CD8+ T-lymphocytes (Fig. 2a) and granzyme B-positive T-lymphocytes (Fig. 2b), usually in close association with CMV-positive cells and apoptotic cells. Some apoptotic cells were also cytomegalic. Diffuse infiltration of scattered CD8+ T-lymphocytes was also detected in the cerebral parenchyma. As in uninfected fetuses, only a few scattered CD4+ T-lymphocytes and CD20+ B-lymphocytes were observed.

Figure 1.

 Laminar necrosis in the third cortical layer (between the two arrows). The cortex has an irregular contour with deepening of the immature sulci, suggestive of early polymicrogyria. Haematoxylin and eosin, ×10.

Figure 2.

 Cerebral white matter. Immunohistochemistry, ×40. (a) A microglial nodule with a cytomegalic cell (arrow) surrounded by CD8+ T-lymphocytes (CD8 as brown cell marker). (b) A cytomegalic cell (arrow) surrounded by activated (granzyme B-positive) cytotoxic CD8+ T-lymphocytes (granzyme B as a brown marker of cytotoxic activation).

The necrosis of cerebral tissue was more frequently associated with inflammatory, apoptotic and CMV-positive cells, although necrotic areas without viral inclusions were also observed. These areas contained abundant macrophages and activated microglia. Gliosis was frequently diffuse, involving all of the white matter, with more pronounced ramified astrocytes in the necrotic and inflamed areas.

Placenta: CMV-positive cells such as fibroblasts and endothelial cells were mainly found in the stromal component of the villi. Only 20% of villi showed normal morphology according to gestational age. Most of the remaining villi (80%) were mainly hydropic, hydropic with focal necrosis, non-hydropic with focal necrosis, or completely necrotic. In completely or partially necrotic villi, the areas of necrosis always showed cytomegalic cells and a prominent surrounding inflammatory infiltrate. Cytomegalic cells were also found within areas with inflammatory cells, but no signs of necrosis.

The inflammatory infiltrate was mainly composed of multifocal aggregates of CD8+ T-lymphocytes, mainly located in the areas with CMV-positive cells and necrosis. Diffuse and scattered CD8+ T-lymphocytes were also identified (chronic villitis). Most of the CD8+ T-lymphocytes were locally activated (granzyme B-positive). Caspase-3-positive cells undergoing apoptosis were often located in close proximity to activated cytotoxic lymphocytes. Some of the apoptotic cells were cytomegalic. Low numbers of CD4+ T-lymphocytes and CD20+ B-lymphocytes were scattered throughout the tissue, similarly to what was observed in uninfected fetuses.

In the other organs, the inflammatory infiltrate was mainly composed of focal aggregates of CD8+ T-lymphocytes, often activated, associated with CMV-positive cells. Rare apoptotic cells were found in the lung and kidney close to CMV-infected cells.

Haematopoiesis.  Areas of extramedullary haematopoiesis were observed in the kidney, where it is not normally present. Moreover, the extent of regular haematopoiesis in liver and pancreas was increased as compared with control cases. Several circulating erythroblasts were also identified in fetal and placental vessels.

Viral loads.  CMV loads in the different tissues are shown in Fig. 3. The highest viral load was in the placenta (mean: 12 700 copies/5 ng DNA), followed by the pancreas (mean: 2400 copies/5 ng DNA), liver, kidney, brain, and lung. The lowest viral load (mean: 57 copies/5 ng DNA) was in the heart.

Figure 3.

 Tissue viral load in fetuses with different degrees of brain damage. The bars show the mean viral load values and the interquartile range for each group.

Ultrasonographic findings.  Table 1 summarizes the results of ultrasound examinations. Ultrasound always gave pathological findings in the seven fetuses with severe histological brain damage, and in six of them the abnormalities observed were in the brain.

Table 1.  Correlation between brain histological findings and ultrasound findings in the 45 cytomegalovirus (CMV)-infected fetuses
GroupNo. of casesBrain histological examinationImmunohistochemistry for CMV in the brainUltrasound findings (no. of cases)
  1. IUGR, intrauterine growth restriction.

A7Severe damagePositiveHyperechogenic bowel, hepatomegaly (1)
Periventricular hyperechogenicity (2)
Ventriculomegaly, hyperechogenic bowel (1)
Ventriculomegaly, periventricular hyperechogenicity, hyperechogenic bowel (1)
IUGR, ventriculomegaly, hydrocephalus, hyperechogenic bowel, hepatomegaly (1)
IUGR, periventricular hyperechogenicity, hyperechogenic bowel (1)
B15Moderate damagePositiveNormal (9)
Hyperechogenic bowel (2)
Periventricular hyperechogenicity (2)
Periventricular hyperechogenicity, hyperechogenic bowel (2)
C6Mild damagePositiveNormal (6)
17No damageNegativeNormal (17)

Group B: 15 fetuses with moderate cerebral damage (score 16–24)

Necrosis in the brain was only focal and associated with cytomegalic cells. Microglial nodules were few, and the inflammatory response was moderate (score 4–8). In the parietal white matter, vessels tended to be increased in number, and to be ectasic, with plump endothelium. Microglia was often abundant, and gliosis was more prominent. Glial karyorrhexis was also seen.

Approximately 50% of the placental villi showed normal morphology, whereas the remaining 50% comprised hydropic or partially necrotic villi. In these cases, placental necrosis was only focal, and did not involve the whole villous. However, necrotic areas always included cytomegalic cells and a prominent inflammatory infiltrate. The infiltrate was also present in areas with CMV-positive cells, but lacking necrosis.

The inflammatory infiltrate was quantitatively less abundant in the placenta and in the other fetal organs than that observed in fetuses with severe brain damage. The immune cell population was generally composed of the same elements and with the same pattern of infiltration, but with fewer aggregates.

Mild extramedullary haematopoiesis was observed in the kidney. Few circulating erythroblasts were identified.

Both the inflammatory infiltrate and the presence of apoptotic cells were the same as in the severe cases, but quantitatively less abundant.

CMV loads in the different tissues are shown in Fig. 3. The tissue viral load of these fetuses was lower than that of fetuses with severe cerebral damage. The sum of all tissue viral loads for each group was considered, and the difference observed was statistically significant (p <0.005). Also in these fetuses, the highest viral load was in the placenta.

Ultrasound in these fetuses gave normal findings in nine cases and abnormal findings in six cases, with pathological neurosonographic findings in four of them (Table 1).

Group C: six fetuses with minor cerebral damage (score 0–15)

Brain histology showed subtle injuries such as white matter oedema, increased number of small vessels with plump endothelium, rare karyorrhectic glial bodies, and focal microglia activation. Gliosis was less pronounced, and the astrocytes were less ramified. This damage was more evident in the parietal white matter.

Approximately 80% of the placental villi showed normal morphology, whereas the remaining 20% of the placenta was composed of scattered hydropic villi.

The inflammatory cells in the placenta and in the other fetal organs were scattered, with only a few aggregates of CD8+ T-lymphocytes, rarely activated, that were often associated with CMV-positive cells (score 0–3).

No difference was observed in extramedullary haemopoiesis as compared with control cases. No circulating erythroblasts were identified, and no apoptotic cells were detected.

CMV loads in the different tissues are shown in Fig. 3. The tissue viral load of these fetuses was lower than that of fetuses with moderate cerebral damage. The sum of the viral loads in all tissues for each group was considered, and the difference was statistically significant (p <0.05). Ultrasound gave no pathological findings (Table 1).

Discussion

Congenital fetal CMV infection is often associated with diffuse CMV positivity detected by immunohistochemistry in several organs [6]. However, the presence of viral antigens in fetal tissue does not necessarily reflect organ damage in newborns. Among infected organs, the brain is well known to have the most severe histological damage [5,6], but the role of a congenitally acquired viral infection in causing brain pathology remains largely unknown.

The present study investigated 45 fetuses from women primarily infected by CMV with a positive prenatal diagnosis. A disseminated CMV infection was observed in all of the organs examined, involving a variety of cell types. The predominant target cells of CMV infection were epithelial and endothelial cells, and this probably explains the more frequent CMV infection in organs such as the pancreas than in the heart, which comprises myocytes of mesenchymal origin. We focused on the effect of CMV infection on the brain, because, unlike other organ injury, brain damage is generally believed to be irreversible.

According to the immunohistochemical results, 38% of the CMV-infected fetuses had CMV-negative brains and did not show any histological brain damage. This is a picture of what is happening at 21 weeks of gestation, and we do not know what may happen later, but it is in agreement with literature reports that high viral loads in amniotic fluid may be associated with symptomatic or asymptomatic congenital infection [9].

On the other hand, 62% of the fetuses had CMV-positive brain tissue, and showed different varieties and degrees of damage. As expected, brain and placenta tissue damage correlated with viral tissue load, suggesting a relationship between the severity of infection and histological lesions.

The highest viral load was found in the fetuses with the most severe brain damage (group A), suggesting a massive and widespread viral infection. The brains of these fetuses showed severe cerebral damage with microglial nodules, cortical and white matter necrosis, and polymicrogyria. Notably, microglial nodules included cytomegalic cells surrounded by activated CD8+ T-lymphocytes and apoptotic cells, suggesting immune-mediated damage.

CD8+ cytotoxic T-lymphocytes are extremely important in the defence against viruses [14]. These effectors are also substantially involved in immune surveillance of the CNS, where they activate and recruit microglial cells [15,16]. However, the presence of CD8+ cytotoxic T-lymphocytes in cerebral tissues may also produce direct injury as a consequence of immune-mediated damage. This mechanism has already been described in HIV-1 encephalitis as a productive immune response against the virus, in which granzyme B-positive CD8+ T-cytotoxic cells usually interact with and kill infected neurons, thereby mediating brain injury [17].

Prominent infiltration of fetal activated CD8+ T-cells was observed in our fetuses with severe brain damage, probably reflecting the occurrence of CMV-specific immune responses. The antigen specificity of these cytotoxic lymphocytes in human fetuses and their role in inducing damage to the brain parenchyma need to be further elucidated.

Group A fetuses showed the most severe cerebral lesions. However, some of the same injuries can be found in hypoxic–ischaemic encephalopathy [18]. Cortical necrosis was observed in the third layer, the most metabolically active region of the cortex. This finding may suggest superimposed hypoxic damage. It is of note that, in the same fetuses, 80% of the parenchyma showed diffuse chronic villitis with necrosis and hydrops. We considered not only non-functional villi showing necrosis with inflammatory infiltrate, but also those with hydrops, because stromal oedema reduces local vascular flow [19]. The same placentas also had a very high viral load (median value: 12 700 copies/5 ng DNA), suggesting that severe placental infection may cause severe tissue damage leading to impaired placental function with reduced oxygenation of the developing fetus. As a consequence, the brain, which is highly sensitive to hypoxia, may be severely injured. Fetal hypoxia in our cases was also suggested by the detection of markedly increased extramedullary haemopoiesis in the liver, kidney, and pancreas. Physiologically, extramedullary haemopoiesis in the fetus is present in the liver, and consists of haemopoietic precursors outside of the bone marrow. However, in response to chronic hypoxaemia, haemopoiesis may increase not only in the liver, but also in other organs, such as the kidney and pancreas [20–23]. Moreover, the detection of circulating erythroblasts in fetal and placental vessels also suggests chronic hypoxia.

The fetuses with severe brain damage showed a prevalence of immune-mediated damage also in other organs. Furthermore, chronic villitis, interstitial pneumonia and chronic pancreatitis were also detected in these cases.

Fifteen fetuses showed moderate cerebral damage (group B), with microglial nodules associated with cytomegalic cells. Brain necrosis, when present, was only focal. Therefore, the tissue damage here was mainly immune-related. However, especially in the parietal white matter, there was evidence of more subtle damage, with vascular changes and glial apoptotic bodies. In the corresponding placentas, the residual functional area was approximately 50%. The scant placental impairment with minor hypoxic fetal injury was also confirmed by a less pronounced increase in extramedullary haemopoiesis and circulating erythroblasts than in the more severe cases. The tissue viral load was significantly lower (p < 0.005) than that in fetuses with severe brain damage.

Six fetuses with minor cerebral damage (group C) showed only telencephalic leukencephalopathy on a background of scattered cytomegalic and inflammatory cells. Telencephalic leukencephalopathy is a very subtle white matter lesion, and a correlation with neurological symptoms can only be supposed [24,25]. These cases had a significantly lower viral load than group A (p < 0.001) and group B (p < 0.05).

The moderate and severe histological brain damage found in 22 (group A and group B) of the 45 brains studied can be considered to be permanent lesions. Only in these cases did we observe abnormal ultrasound findings, in 13 fetuses. Nine fetuses showed hyperechogenic bowel, which was associated with ultrasound brain abnormalities in six cases, hepatomegaly in one case, and isolated findings in two cases. The significance of hyperechogenic bowel in pregnancy remains unsettled [26–28]. However, our three fetuses with hyperechogenic bowel without ultrasound brain abnormalities at 21 weeks of gestation but with moderate and severe histological brain damage presented a high CMV load in the amniotic fluid. The combination of such a non-specific ultrasound finding with a viral load of >105 copies/mL in the amniotic fluid should be taken into consideration when patients are counselled.

Moreover, ultrasound at 21 weeks of gestation gave normal findings in 32 cases and, among these fetuses, histological brain damage (mild and moderate) was detected in 15. Therefore, even when associated with normal ultrasound findings, a high viral load in the amniotic fluid was predictive of histological brain damage in 47% of our population. This information should also be taken into consideration when patients are counselled.

The placenta seems to be the most critical target of congenital CMV infection, as there is a correlation between the viral load in the placenta and that in other fetal organs. When a low viral load was found in the placenta (median value: 33 copies/5 ng DNA), a low viral load was also seen in the other fetal tissues. In these cases, the inflammatory response was limited, and no immune-mediated organ damage was observed.

When the placental CMV load was higher (median value: 5490 copies/5 ng DNA), the viral load in the other fetal tissues was also higher. In these cases, major infiltration of activated cytotoxic CD8+ T-lymphocytes, presumably responsible for immune-mediated damage, was mainly observed in placenta and brain tissue.

Furthermore, in the presence of a severe placental infection (median value: 12 700 copies/5 ng DNA) with diffuse villitis and necrosis, placental function is very probably compromised, causing fetal hypoxia. The tissue viral load, inflammatory infiltrate and immune-mediated damage were much more prominent in these fetuses. Therefore, the tissue viral loads in the placenta and brain closely correlate with the histological damage: the more virus that is detected in the tissues, the greater the immune response and the greater the damage.

CNS anomalies induced by congenital CMV infection are likely to result from a direct effect of viral replication in the brain and an indirect effect occurring at two different levels: in the brain, where the infection can induce immune-mediated damage, and in the placenta, where the infection can cause placental insufficiency and, consequently, hypoxic brain damage.

Acknowledgements

This work was supported by grants (SA43GABR) from St Orsola-Malpighi University Hospital Bologna, as the prize for winning a hospital competition for the presentation of scientific research projects.

Transparency Declaration

The authors have no conflicts of interest to declare.

Ancillary