Recorded sessions on demand will be available 24 hours after the session ends

Abstract Body

Cytomegalovirus (CMV) is the most common congenital infection. Congenital CMV (cCMV) is an important cause of childhood hearing loss (HL) and neurodevelopmental disability. Clinical consequences of cCMV range from asymptomatic infection without sequelae to lifelong disabilities or death. Evaluation of central nervous system (CNS) involvement is crucial in deciding on antiviral treatment, planning follow-up and early intervention, as well as counseling parents regarding prognosis.

A review and new data on biomarkers of brain involvement in cCMVwill be provided.

Absolute microcephaly (head circumference Z-score – birth weight Z-score <–2) and chorioretinitis are clinical predictors of poor outcome.

Low viral load is associated with a low risk of HL. However, the positive predictive value of viral load for predicting HL is low. CMV DNA detection in cerebrospinal fluid (CSF) is associated with poor outcomes. Still, infants without active encephalitis at birth but with brain abnormalities owing to early fetal CMV CNS infection are also at risk for neurologic sequelae.

Neonatal neuroimaging (i.e. a combination of ultrasonography and MRI) predicts outcome in cCMV. Disruptive lesions and cerebral dysgenesis correlate with a poor prognosis. In infants with isolated white matter injury, its extent is helpful in outcome prediction.

Elevated CSF protein and β2-microglobulin (a marker of immune activation and increased cellular turnover) have prognostic significance in newborns with cCMV. Other CSF biomarkers may inform of the mechanisms and severity of brain injury.

Consideration of the severity of neuroimaging abnormalities in combination with other biomarkers of brain involvement maximizes predictive ability in neonates with cCMV.

Abstract Body

Introduction: Prompt and accurate diagnosis, leading to adequate use of antibiotics is key to decrease late onset sepsis (LOS) related morbidity and mortality.

Objective: to test the diagnostic value of heart rate variability integrating new visibility graph indexes in machine learning algorithms to diagnose LOS.

Materials and Methods: After parental consent infants born before 30 weeks gestation (2017-18, 6 NICU) have been studied as part of the Digi-NewB cohort (NCT02863978, EU GA n°689260). The LOS group consisted in patients who received more than 5 days antibiotics beginning after 72 hours of life. The control group consisted in infants who didn’t receive antibiotics. HRV features were calculated prior to the start of antibiotic (LOS group) or in a randomly selected period (control group). After an automatic feature selection (U test and principal component analysis) we trained 4 machine learning algorithms using leave one out cross-validation and area under the receiver operating characteristics curve (AUROC).

Results: Patients in the LOS group [n=24, GA=26.5 (25.3-28)weeks; BW=840 (740-1025)] were compared to the control group [n=25, GA=28 (27-28.5) weeks; BW=1107 (925-1260)]. LOS occurred at 8.4 (5.6-10.5) days. Most of HRV parameters studied were informative. After feature transformation the best performance was obtained by logistic regression algorithm integrating visibility graph indexes using a 48h-calibration with 88% and 85% AUROC during the 6 and 12 hours preceding antibiotics.

Conclusion: These results, based on machine learning algorithms with introduction of visibility graph index alongside the traditional HRV features, suggest the possibility of non-invasive monitoring of the risk of LOS.

Abstract Body

Background and Aim

Early-onset sepsis (EOS) is a life-threatening event associated with serious morbidity and adverse neurodevelopmental outcome. Limited data are available regarding the role of fetal therapy (FT) as a potential cause of EOS. We aimed to evaluate long-term outcome in very low birth weight (VLBW) infants in relation to FT and EOS.

Methods

Neurodevelopment was assessed in VLBW survivors at 2 years of corrected age using Bayley Scales of Infant and Toddler Development 3^rd Edition. Clinical data were obtained retrospectively from medical records.

Results

434 VLBW infants were included in the analysis (mean gestational age 28.8 ± 2.6 weeks, birth weight 1087 ± 281 grams). EOS was diagnosed in 8.8% of patients and 18% of EOS patients underwent FT. We observed significant association between FT and EOS (p = 0.037). Furthermore, both variables were independent predictors for adverse long-term neurodevelopmental outcome. Fetal therapy was significantly associated with worse cognitive (p = 0.004), language (p = 0.020) and motor (p < 0.001) skills on the Bayley assessment. The association was also significant with cerebral palsy (p < 0.001). Similarly, we observed significant correlation between EOS and adverse neurodevelopment in terms of cognitive (p < 0.001), language (p = 0.013) and motor (p = 0.016) skills.

Conclusions

We found strong relationship between fetal therapy and EOS that represents a significant risk factor for adverse neurodevelopment at 2 years of corrected age in VLBW infants.

The project was supported by the Research Project NV 17-31403A.