During current clinical activity we prospectively collected all clinical and laboratory data relating to newborns hospitalized with defined enterovirus meningeal infection and followed up 12 months at the outpatient service of Bambino Gesù Children's Hospital (Rome, Italy) from 2015 to 2021. Data of patients were obtained from medical records.

According to our neonatal department protocols, all infants admitted to the Neonatal Intensive Care Unit (NICU) with fever undergo contact isolation and laboratory sepsis work-up: complete blood counts, blood chemistry tests, blood culture, urinalysis, urine culture, C-reactive protein (CRP), procalcitonin (PCT), rectal swab for multidrug-resistant bacteria, nasal swab for Staphylococcus aureus. The lumbar tap is routinely performed as part of the initial sepsis work-up in all febrile neonates and infants. We examine the cerebrospinal fluid (CSF) samples for routine cell counts and chemistry and culture. We also performed real-time polymerase chain reaction (RT-PCR) on CSF, to detect the genome of viruses that most frequently cause meningoencephalitis in the newborn, such as herpes virus 1–2–6, cytomegalovirus, enterovirus and parechovirus.

Viral meningitis was defined on the basis of two or more of the following clinical signs (fever > or = 38°C, irritability, crying exhaustion, skin marbling, presence of skin rash, abdominal distension, diarrhea, poor feeding, bulging of anterior fontanel), associated with RT-PCR positivity of CSF and/or blood for RNA enteroviruses, and a CSF culture negative for bacteria. The clinical presentation of parechovirus central nervous system (CNS) infection is similar to that of enteroviral CNS infection, but with RT-PCR positivity of CSF and/or blood for parechovirus.

Although CSF pleocytosis is commonly absent in infants under 30 days of age in EV-positive infants (19), its presence may be observed in about half of cases of EV meningitis (20). Therefore, we also collected data on the CSF pleocytosis.

Nucleic acids were extracted from 400 μl of CSF and plasma samples, using the automatic platform QIAsymphony (Qiagen) and the Virus/Pathogen Midi kit (Qiagen), according to manufacturer's instructions. RT-PCR was performed on 7,900 HT Real Time PCR System (Applied Biosystems) with Enterovirus R-gene Argene (Biomérieux) or Parechovirus R-gene (Biomérieux), targeting the highly conserved sequences of the terminal 5' non-coding region of Enteroviruses or Parechoviruses, according to manufacturer's instructions.

Enterovirus RNA positive samples, after the extraction step, were submitted to sequencing analysis of part of the VP1 genomic region, coding for one of the capsid proteins, according to WHO recommendations (21).

The targeted region of the viral RNA sequence was reverse transcribed to single stranded complementary DNA (cDNA) via primer extension with the Multiscribe Reverse Transcriptase enzyme (Thermofisher Scientific) on Gene Amp PCR System 9,700 (Applied Biosystem), using four different primers.

Amplifications were carried on with AmpliTaq GoldTM DNA Polymerases (Applied Biosystem) on Gene Amp PCR System 9,700, using two sets of primers.

Sequencing reaction was set on with AN89 and AN88 primers and the Big Dye terminator v3.1 ready reaction mix (Life Technologies). Sequencing analysis was carried on ABI PRISM 3130 XL Genetic Analyzer (Applied Biosystem) according to manufacturer's conditions.

The sequences obtained were then compared with the published prototype sequences to identify unknown enteroviruses through comparison of partial VP1 sequence data (22).

In our Neonatal Intensive Care Unit all infants with meningitis of any etiology are monitored with amplitude-integrated electroencephalogram (aEEG) in the first 48–72 h of infection. Then the monitoring is suspended if the newborn improves and there are no suspected tracing anomalies. In case of suspected anomalies of the aEEG, or if the aEEG instrument is not available, an electroencephalogram (EEG) is performed, to rule out seizures. Immediately after carrying out the microbiological cultures, we start empiric antibiotics. If the diagnosis is of enteroviral meningitis, we suspend antibiotics and start immunoglobulins intravenously for three days, according to our local protocol. All of the infants described in this report followed this path of diagnosis and therapy.

In our Unit all neonates with enteroviral meningitis undergo brain ultrasound (US) during the hospitalization. In newborns enrolled in this study the MRI examination was performed only in 26/30 (86.7%) newborns, as the protocol for the management of viral meningitis has changed over time. In the past MRI on neonates with viral meningitis was performed as a second level exam, if the brain US examination was pathological. The introduction of the mattress technique for brain MRI currently makes it possible to extend the examination to newborns with normal US, to study them better. The mattress technique allows to avoid sevoflurane sedation, that we administer in order to acquire MRI images with a good quality, only in the case of the child's restlessness. To date we did not observe any side effects. For this procedure we always require the informed consent of the patients' parents. MRI exam is generally performed within 7–15 days after the onset of the infection by 3 T scanner (MAGNETOM Skyra, Siemens, Erlangen, Germany).

After discharge from the NICU, children enter in an outpatient multidisciplinary follow-up, with quarterly clinical and instrumental assessments up to 12 months of age. All infants of this cohort were regularly evaluated to early intercept the presence of a neurological disability: the presence of a cerebral palsy (CP) was defined according to Bax et al. (23). To evaluate neurodevelopmental outcomes, we used the Bayley Development Scale for Toddlers and Infants Third Edition (BSDI-III, 2006) (24). This scale provided scores for three major development domains: cognition, language, and motor. Scores between 85 and 115 indicate normal development, while scores below 85 (-1 SD) indicate a developmental delay in the domain evaluated. The assessments were administered to patients by a developmental psychologist (S.B.), trained in BSID test procedures. The examiner was blinded to the patients' MRI findings and neonatal course. The psychologist assessed neurodevelopmental outcomes until 12 months. Children with scores within the normal range in all three domains were considered normal; children with a score below 85 (-1 SD) in at least one of the three domains were considered affected by a neurodevelopmental delay (25).

Visual function and retinal diseases were assessed by repeated examinations of the fundus and functional tests (electroretinogram–ERG—and visual evoked potentials –VEP—when necessary). Normal vision was defined as the “absence of any detectable pathology of the visual system”; mild abnormal vision as “the presence of a mild impairment which allowed useful vision”, and severe visual impairment as “a child functionally blind or perceives light only” (26).

Hearing function was explored by brainstem auditory evoked potentials (BAEPs). Auditory global function was defined as normal in “absence of any detectable pathology”, as mild if requiring hearing aids, or as severe if functionally deaf (uncorrected even with aids) (27).

Data are shown as numbers and percentages for categorical variables, whereas continuous variables are expressed as mean ± standard deviation (SD). Pearson correlations were performed to investigate the relationship between the results of the Bayley III composite score (cognitive, language, and motor) and the clinical characteristics of the infants' diagnosis at admission. Neurodevelopmental assessment in children with normal and pathological MRI were compared using ANOVA. Multiple logistic regression analysis was used to determine the most important predictors of pathological MRI. The odds ratio and 95% confidence interval for each variable were calculated. A p < 0.05 was considered statistically significant. Statistical analysis was performed using software SPSS (version 20 for Windows).

We longitudinally observed 30 children, full-term born, who contracted enterovirus and parechovirus meningitis within the first month of life (mean age at diagnosis was 15.80 ± 7.33 days). Mean gestational age was 39.7 ± 0.8 weeks and mean birthweight was 3,457 ± 405 grams. Viral genome typing was performed on 15/30 (50.0%) neonates. Echoviruses (7, 9, 11, 18, 20 and 30) and coxsackie viruses (A9 and B5) RNA were detected in 12/15 cerebrospinal fluid samples while parechovirus RNA on 3/15 samples. Two of those three babies with parechovirus meningitis showed abnormal brain MRI.

The Table 1 shows main clinical characteristics of our population of neonates and their length of hospital stay.

The most frequent symptom at the onset of the infection (100% of neonates) was high fever, even up to 39.2°C, associated with marked irritability during examination procedures and poor feeding.

All infants underwent aEEG monitoring, without seizure finding. Seizures were also ruled out by EEG in a total of seventeen infants (54.8%) with dubious aEEG trace.

One of those children with a human herpesvirus 6 (HHV6) coinfection, had epileptic anomalies on the electroencephalographic trace, in particular low waves in the parieto-temporal site and a poor background organization. The diagnosis of epilepsy was never made by neurologists, the EEG trace progressively improved, and anticonvulsant therapy was never necessary. Currently this child is fine and presents a normal neurodevelopmental outcome. Due to the presence of HHV6 co-infection we started therapy with intravenous Ganciclovir and then with oral Valganciclovir.

In summary, 29/30 patients (96.7%) had EEG and/or aEEG trace with no signs of seizures.

In 55% of our infants we found an alteration of the CSF. Of these, 32% had frankly pathological liquor, with a predominance of mononuclear cells. The three infants with parechovirus meningitis had CSF with less than 5 white blood cells and normal levels of protein and glucose. Pleocytosis was not associated with a more serious course of acute disease nor with worse long-term outcomes. Table 2 shows the CSF characteristics of our patients.

In all 30 enrolled patients we identified viral RNA. We were able to perform the genetic typing only on 15/30 (50.0%) CSF samples. In 9 of them we typed echoviruses (7,9,18,11,20,30): echovirus 30 in three and echovirus 18 in two samples; the other virus typed were coxsackie B5 in one patient's CSF and coxsackie A9 in two patients' CSF. Three infants had parechovirus meningitis (Supplementary Table 1). One infant with enterovirus meningitis had HHV6 co-infection. The absence of viral DNA determined with the analysis of the child's hair bulb allowed us to exclude the chromosomal integration of the viral genome.

We did not isolate EV71, which seems to be the most neurotropic and the one most frequently involved in the genesis of brainstem encephalitis (28).

In our patient with brainstem lesions, we isolated echovirus 30, but such lesions were considered secondary to perinatal asphyxia, according to the radiological characteristics and the history of the patient.

All 30 patients underwent brain ultrasound (US), while only 26 (86.7%) also performed brain MRI. The NICU protocol concerning viral meningitis has been modified over time. At the beginning the brain MRI was performed only in newborns with brain US abnormalities. Subsequently we decided to carry out the examination to all patients, having introduced the heated mattress technique, which avoids the sedation of the baby. We did not detect abnormalities in all newborns by brain ultrasounds. Conversely, we found non-specific brain abnormalities in 9/26 newborns (34.6%) by, which was performed on patients between 12 and 65 days of life (mean 38 ± 26 days). Among 9 neonates with MRI anomalies, dural contrast enhancement was found in 3/9 (33.3%) neonates (one of them with echovirus 20 infection); diffuse white matter alterations, probably due to delay of myelination, in 2/9 (22.2%) neonates (one had an infection by echovirus 7); cytotoxic oedema in 1/9 (11.1%) and periventricular microbleeds in 1/9 (11.1%) neonates (infected by coxsackievirus B5). Moreover, 1/9 (11.1%) neonate with echovirus 11 infection presented brainstem alterations and 1/9 (11.1%) with echovirus 30 showed mildly hyper-intense signals on T2-weighted images in the head of right caudate nucleus. Among the 17 patients with normal MRI one presented a subdural effusion, perhaps due to a difficult delivery, but the MRI was considered normal by neuro-radiologists. Analyzing data by logistic regression model (Table 3), pathological brain MRI was associated with age at admission to the NICU: the younger the infant, the greater the likelihood of brain MRI abnormalities (OR: 0.788; 95% CI: 0.634 −0.978; p = 0.031).

During the clinical follow-up, children displayed normal weight growth rates. The neurodevelopmental assessment by Bayley-III is presented in Table 4. At 12 months of life we found that most babies (96.7%) had no sequelae on cognitive, language, and motor BSDI-III composite score results. None had a cerebral palsy. Only one child, with previous perinatal asphyxia, had a mild delay in fine and gross motor skills and in receptive language. All infants reported to have a normal vision and a normal hearing function.

Correlations between Bayley III composite score (cognitive, language, and motor) and the clinical characteristics of the infants' diagnosis at admission are reported in Supplementary Table 2.

Children with pathological MRI during hospitalization scored significantly lower than children with normal MRI on cognitive Bayley III subscale at Neurodevelopmental assessment at one year of age, even if all the scores obtained fell within the normal range (F = 6.289; p = 0.022) (Table 5).