Cerebral palsy after very preterm birth - an imaging perspective

Fetal Intraparenchymal Hemorrhage Imaging Patterns, Etiology, and Outcomes: A Single Center Cohort Study

OBJECTIVE

This study examines associations among fetal brain magnetic resonance imaging (MRI) injury patterns, etiologies, and outcomes in fetal intraparenchymal hemorrhage (IPH).

METHODS

This is a retrospective, single-center cohort study of IPH diagnosed on fetal MRI (1996-2022). IPH and associated abnormalities were categorized by 2 pediatric neuroradiologists; electronic medical records were reviewed by 2 pediatric neurologists to classify etiology and outcomes including cerebral palsy, epilepsy, developmental delay, and death.

RESULTS

Forty-four fetuses with IPH were identified (34 singleton and 10 twin gestations) with MRI at median 24 weeks gestation (interquartile range [IQR] = 22-28 weeks). IPH was commonly supratentorial (84%) and focal (50%) or focal with diffuse injury (43%) and was often associated with germinal matrix hemorrhage (GMH; 75%) and/or intraventricular hemorrhage (IVH; 52%). An etiology was identified in 75%, including twin-twin transfusion syndrome (TTTS, n = 10), COL4A1/2 variants (n = 8), or other fetal/maternal conditions (n = 15). COL4A1/2 variants were associated with focal IPH and the presence of hemorrhagic porencephaly, and intrauterine transfusion was associated with infratentorial hemorrhage. Twenty-two fetuses were liveborn, and 18 pregnancies were terminated. Among those with follow-up ≥ 12 months (median = 7 years), 12 of 13 had cerebral palsy, 6 of 13 had developmental delay, and 5 of 13 had epilepsy.

INTERPRETATION

An etiology for fetal IPH with or without GMH-IVH is identified in most cases in our cohort and is commonly TTTS, COL4A1/2 variants, or other maternal/fetal comorbidities. Pattern of fetal IPH on MRI is associated with etiology. Cerebral palsy and neurodevelopmental impairment were common in liveborn infants. Genetic studies should be considered in cases of fetal IPH without an otherwise apparent cause. ANN NEUROL 2024.

Fibrinogen inhibits sonic hedgehog signaling and impairs neonatal cerebellar development after blood-brain barrier disruption

Cerebellar injury in preterm infants with central nervous system (CNS) hemorrhage results in lasting neurological deficits and an increased risk of autism. The impact of blood-induced pathways on cerebellar development remains largely unknown, so no specific treatments have been developed to counteract the harmful effects of blood after neurovascular damage in preterm infants. Here, we show that fibrinogen, a blood-clotting protein, plays a central role in impairing neonatal cerebellar development. Longitudinal MRI of preterm infants revealed that cerebellar bleeds were the most critical factor associated with poor cerebellar growth. Using inflammatory and hemorrhagic mouse models of neonatal cerebellar injury, we found that fibrinogen increased innate immune activation and impeded neurogenesis in the developing cerebellum. Fibrinogen inhibited sonic hedgehog (SHH) signaling, the main mitogenic pathway in cerebellar granule neuron progenitors (CGNPs), and was sufficient to disrupt cerebellar growth. Genetic fibrinogen depletion attenuated neuroinflammation, promoted CGNP proliferation, and preserved normal cerebellar development after neurovascular damage. Our findings suggest that fibrinogen alters the balance of SHH signaling in the neurovascular niche and may serve as a therapeutic target to mitigate developmental brain injury after CNS hemorrhage.

Spectrum of Fetal Intraparenchymal Hemorrhage in COL4A1/A2-Related Disorders

BACKGROUND

COL4A1/A2 variants affecting the alpha 1 and 2 chains of type IV collagen are increasingly recognized as a cause of fetal and neonatal intracranial hemorrhage, porencephaly, and schizencephaly. Fetal magnetic resonance imaging (MRI) findings in COL4A1/A2-related disorders are not well characterized.

METHODS

This is a retrospective case series of fetal MRI findings in eight patients with intraparenchymal hemorrhage (IPH) and COL4A1/A2 variants, five of whom have postnatal imaging and clinical follow-up.

RESULTS

IPH was multifocal and bilateral in four of eight patients. IPH involved the frontal lobes in all cases and basal ganglia in six of eight. The median maximum diameter of IPH was 16 mm (range 6 to 65 mm). All patients had ventriculomegaly, and four of eight had intraventricular hemorrhage. Prenatal IPH size correlated clinically with motor outcomes, and none had clinically symptomatic recurrent hemorrhage.

CONCLUSION

COL4A1/A2 variants can present with a spectrum of IPH prenatally, including small and/or unifocal IPH, as well as multifocal and bilateral IPH, involving the frontal lobes and basal ganglia. Given the wide spectrum of IPH severity seen on fetal brain MRI, genetic testing for COL4A1/A2 variants should be considered in all cases of fetal IPH.

A multi-task, multi-stage deep transfer learning model for early prediction of neurodevelopment in very preterm infants

Survivors following very premature birth (i.e., ≤ 32 weeks gestational age) remain at high risk for neurodevelopmental impairments. Recent advances in deep learning techniques have made it possible to aid the early diagnosis and prognosis of neurodevelopmental deficits. Deep learning models typically require training on large datasets, and unfortunately, large neuroimaging datasets with clinical outcome annotations are typically limited, especially in neonates. Transfer learning represents an important step to solve the fundamental problem of insufficient training data in deep learning. In this work, we developed a multi-task, multi-stage deep transfer learning framework using the fusion of brain connectome and clinical data for early joint prediction of multiple abnormal neurodevelopmental (cognitive, language and motor) outcomes at 2 years corrected age in very preterm infants. The proposed framework maximizes the value of both available annotated and non-annotated data in model training by performing both supervised and unsupervised learning. We first pre-trained a deep neural network prototype in a supervised fashion using 884 older children and adult subjects, and then re-trained this prototype using 291 neonatal subjects without supervision. Finally, we fine-tuned and validated the pre-trained model using 33 preterm infants. Our proposed model identified very preterm infants at high-risk for cognitive, language, and motor deficits at 2 years corrected age with an area under the receiver operating characteristic curve of 0.86, 0.66 and 0.84, respectively. Employing such a deep learning model, once externally validated, may facilitate risk stratification at term-equivalent age for early identification of long-term neurodevelopmental deficits and targeted early interventions to improve clinical outcomes in very preterm infants.