基于遗传负荷构建新生儿重症监护室轻度脑电异常患儿预后不良预测模型的巢式病例对照研究OA

Background Mild electroencephalogram(EEG)abnormalities are extremely common in the Neonatal Intensive Care Unit(NICU).However,their prognoses vary significantly—while some cases represent transient changes,others insidiously progress into severe neurodevelopmental disorders.Objective To investigate the association between genetic risk genes and adverse outcomes in NICU patients with mild EEG abnormalities,and to construct and validate a clinically actionable predictive model.Design A nested case-control study.Methods Patients admitted to the NICU at the Children's Hospital of Fudan University between June 2020 and December 2024 who presented with mild EEG abnormalities and underwent genetic testing were enrolled.The spectrum of genetic disorders among patients with positive genetic diagnoses were analyzed.Children with negative genetic diagnoses were used as the cohort population.Poor prognosis(case group)was defined as the presence of any one of the following at 1 year of age:neurodevelopmental delay,motor developmental delay,cerebral palsy,or death.Otherwise,good prognosis was defined as the outcome of the disease.Patients with favorable outcomes were matched to the case group at a 3∶1 ratio using propensity score matching to serve as the control group.The carrier rates or burdens of rare deleterious variants within specific target genes between the two groups were compared to identify potential risk genes,and a"rare variant burden score"for each patient was calculated.Additionally,quantitative EEG features were extracted and clinical data were collected.The patients included from June 1,2020 to February 1,2024 were used as the model construction set,and the children included from February 2,2024 to December 31,2024 were used as an independent validation set for model validation.Main Outcome Measures The spectrum of genetic disorders and the performance of the predictive model.Results Between June 2020 and February 2024,a total of 1,082 neonates admitted to the NICU with mild EEG abnormalities were enrolled in the study.The cohort comprised 609 males(56.3%)and 473 females(43.7%).Among them,235 cases(21.7%)received a positive genetic diagnosis,exhibiting significant genetic heterogeneity:91 cases involved a total of 103 genes related to brain development and function—predominantly associated with neurological,metabolic,and syndromic disorders—with ABCC8 identified as the most common gene;conversely,153 cases carried 177 genes unrelated to brain development and function—most frequently G6PD—involving genes across multiple systems,including metabolic system,circulatory system and hematological system.There were 847 cases with a negative genetic diagnosis;these constituted the model construction set,comprising 543 cases(including 133 cases in the case group and 410 in the matched control group).The validation set consisted of 80 cases,including 20 cases in the case group and 60 cases in the control group.There were no statistically significant differences in clinical baseline characteristics between the model construction set and the validation set(all P>0.05).Within the model construction set,gene-level burden analysis identified 54 risk genes associated with poor prognosis;furthermore,42 core EEG features were extracted.Based on univariate logistic regression analysis,four clinical phenotypes(clinical seizures,respiratory failure/distress,neonatal encephalopathy/brain injury,and cranial MRI findings)and one genetic predictor(rare variant burden score)were selected for model construction.The combined model,incorporating the four clinical phenotypes,EEG features,and the genetic predictor,yielded an AUC of 0.80(95%CI:0.76-0.83)in the model construction set,and an AUC of 0.86(95%CI:0.74-0.98)in the validation set.Conclusion Genetic factors play a pivotal role in the prognostic stratification of NICU patients with mild EEG abnormalities;furthermore,the proposed genetic burden model represents a promising tool for identifying neonates at high risk of adverse neurodevelopmental outcomes,thereby offering a novel strategy for the realization of precision child neurology.