Publications

Neurodevelopmental impairments associated with congenital heart disease (CHD) may arise from perturbations in brain developmental pathways, including the formation of sulcal patterns. While genetic factors contribute to sulcal features, the association of noncoding de novo variants (ncDNVs) with sulcal patterns in people with CHD remains poorly understood. Leveraging deep learning models, we examined the predicted impact of ncDNVs on gene regulatory signals. Predicted impact was compared between participants with CHD and a jointly called cohort without CHD. We then assessed the relationship of the predicted impact of ncDNVs with their sulcal folding patterns. ncDNVs predicted to increase H3K9me2 modification were associated with larger disruptions in right parietal sulcal patterns in the CHD cohort. Genes predicted to be regulated by these ncDNVs were enriched for functions related to neuronal development. This highlights the potential of deep learning models to generate hypotheses about the role of noncoding variants in brain development.

BACKGROUND AND OBJECTIVES: Hypotonia is a relatively common finding among infants in the neonatal intensive care unit (NICU). Consideration of genetic testing is recommended early in the care of infants with unexplained hypotonia. We aimed to assess the diagnostic yield and overall impact of exome and genome sequencing (ES and GS).

METHODS: Consecutive infants with hypotonia were identified from research and clinical databases across 5 teaching hospitals in United States, Canada, United Kingdom, and Australia. Inclusion criteria included NICU admission and genetic evaluation. Infants with a known explanation for hypotonia were excluded. Data regarding infant characteristics, genetic testing, and diagnoses were collected. The primary outcome was identification of a molecular diagnosis. Impact on care was a secondary outcome. The Fisher exact and Wilcoxon rank-sum tests were used for statistical analysis.

RESULTS: We identified 147 infants with unexplained hypotonia. The median gestational age was 39 weeks (interquartile range [IQR] 36-42 weeks), 77 (52%) were female, and the median age was 8 days at the time of evaluation (IQR 2-19 days). Eighty (54%) had hypotonia as the main clinical feature while 67 (46%) had additional multisystem involvement. Seventy-five (51%) underwent rapid ES, 44 (30%) rapid GS, 2 (1%) both ES and GS, and 26 (18%) were admitted before ES or GS became available. Of the 121 infants who underwent ES and/or GS, 72 (60%) had the primary outcome of a molecular diagnosis. In addition, 2 infants with mitochondrial genome variants were diagnosed by mitochondrial GS after negative ES, and one infant needed targeted testing to identify a short tandem repeat expansion missed by GS. The proportion diagnosed by ES and GS was not different between infants with hypotonia as the primary finding (37/56, 66%) and infants with multisystemic symptoms (35/65, 54%, odds ratio [OR] 1.7, CI 0.8-3.7, p value = 0.20). Testing was more likely to have an impact on care for infants receiving a genetic diagnosis (57/66 vs 14/33, OR 8.4, CI 2.9-26.1, p = 1.0E-05).

DISCUSSION: Rapid ES and GS provided a molecular diagnosis for most of the infants with unexplained hypotonia who underwent testing. Further studies are needed to assess the generalizability of these findings as increased access to genetic testing becomes available.

CLASSIFICATION OF EVIDENCE: This study provides Class IV evidence that in unexplained neonatal hypotonia, rapid ES or GS adds diagnostic specificity.

BACKGROUND: Endocardial fibroelastosis (EFE) is a major effector in the maldevelopment of the heart in patients with congenital heart disease. Despite successful surgical removal, EFE can redevelop, but the underlying cause of EFE recurrence remains unknown. This study aimed to identify hemodynamic predictors and genetic links to epithelial/endothelial-to-mesenchymal transition (EMT/EndMT) alterations for preoperative risk assessment.

METHODS: We assessed the impact of preoperative hemodynamic parameters on EFE recurrence in a cohort of 92 patients with congenital heart disease who underwent left ventricular (LV) EFE resection between January 2010 and March 2021. Additionally, whole-exome sequencing in 18 patients was used to identify rare variants (minor allele frequency <10-5) in high-expression heart (HHE) genes related to cardiac EMT/EndMT and congenital heart disease.

RESULTS: EFE recurred in 55.4% of patients, within a median of 2.2 years postsurgery. Multivariable analysis revealed specific hemodynamic parameters (mitral valve inflow and area, LV filling pressure, and aortic valve gradient and diameter) as predictors, forming a predictive model with an area under the receiver operating characteristic curve of 0.782. Furthermore, 89% of the patients exhibited damaging variants in HHE genes, with 38% linked to cardiac EMT/EndMT Gene Ontology processes and 22% associated with known congenital heart disease genes. Notably, HHE genes associated with cardiac EMT/EndMT were significantly associated with faster EFE recurrence in a multivariate analysis (hazard ratio, 3.56; 95% confidence interval, 1.24-10.17; P = .018).

CONCLUSIONS: These findings established a predictive scoring system using preoperative hemodynamic parameters for EFE recurrence risk assessment. Alterations in HHE genes, particularly those linked to cardiac EMT/EndMT, exacerbate the risk of recurrence.

BACKGROUND: Infant alertness and neurologic changes can reflect life-threatening pathology but are assessed by physical exam, which can be intermittent and subjective. Reliable, continuous methods are needed. We hypothesized that our computer vision method to track movement, pose artificial intelligence (AI), could predict neurologic changes in the neonatal intensive care unit (NICU).

METHODS: We collected video data linked to electroencephalograms (video-EEG) from infants with corrected age less than 1 year at Mount Sinai Hospital in New York City, a level four urban NICU between February 1, 2021 and December 31, 2022. We trained a deep learning pose recognition algorithm on video feeds, labeling 14 anatomic landmarks in 25 frames/infant. We then trained classifiers on anatomic landmarks to predict cerebral dysfunction, diagnosed from EEG readings by an epileptologist, and sedation, defined by the administration of sedative medications.

FINDINGS: We built the largest video-EEG dataset to date (282,301 video minutes, 115 infants) sampled from a diverse patient population. Infant pose was accurately predicted in cross-validation, held-out frames, and held-out infants with respective receiver operating characteristic area under the curves (ROC-AUCs) 0.94, 0.83, 0.89. Median movement increased with age and, after accounting for age, was lower with sedative medications and in infants with cerebral dysfunction (all P < 5 × 10-3, 10,000 permutations). Sedation prediction had high performance on cross-validation, held-out intervals, and held-out infants (ROC-AUCs 0.90, 0.91, 0.87), as did prediction of cerebral dysfunction (ROC-AUCs 0.91, 0.90, 0.76).

INTERPRETATION: We show that pose AI can be applied in an ICU setting and that an EEG diagnosis, cerebral dysfunction, can be predicted from video data alone. Deep learning with pose AI may offer a scalable, minimally invasive method for neuro-telemetry in the NICU.

FUNDING: Friedman Brain Institute Fascitelli Scholar Junior Faculty Grant and Thrasher Research Fund Early Career Award (F.R.). The Clinical and Translational Science Awards (CTSA) grant UL1TR004419 from the National Center for Advancing Translational Sciences. Office of Research Infrastructure of the National Institutes of Health under award number S10OD026880 and S10OD030463.

(1) Background: To identify reasons for the persistence of surgical ligation of the patent ductus arteriosus (PDA) in premature infants after the 2019 Food and Drug Administration (FDA) approval of transcatheter device closure; (2) Methods: We performed a 10-year (2014-2023) single-institution retrospective study of premature infants (<37 weeks) and compared clinical characteristics and neonatal morbidities between neonates that underwent surgical ligation before (epoch 1) and after (epoch 2) FDA approval of transcatheter closure; (3) Results: We identified 120 premature infants that underwent surgical ligation (n = 94 before, n = 26 after FDA approval). Unfavorable PDA morphology, active infection, and recent abdominal pathology were the most common reasons for surgical ligation over device occlusion in epoch 2. There were no differences in demographics, age at closure, or outcomes between infants who received surgical ligation in the two epochs; (4) Conclusions: Despite increasing trends for transcatheter PDA closure in premature infants, surgical ligation persists due to unfavorable ductal morphology, active infection, or abdominal pathology.

OBJECTIVE: Impairments in the maternal-fetal environment are associated with adverse postnatal outcomes among infants with congenital heart disease. Therefore, we sought to investigate placental anomalies as they related to various forms of fetal congenital heart disease (FCHD).

METHODS: We reviewed the placental pathology in singleton pregnancies with and without FCHD. FCHD was divided into separate categories (transposition physiology, obstructive left, obstructive right, biventricular without obstruction, and others). Exclusion criteria included other prenatally known structural malformations and/or aneuploidy. The significance threshold was set at p < 0.05 or False Discovery rate q < 0.05 when multiple tests were performed.

RESULTS: The cohort included 215 FCHD and 122 non-FCHD placentas. FCHD placentas showed increased rates of maternal vascular malperfusion (24% vs. 5%, q < 0.001) and cord anomalies (27% vs. 1%, q < 0.001). Placentas with fetal TGA demonstrated a lower rate of hypoplasia when compared with other FCHD types (1/39 vs. 51/176, Fisher's exact p = 0.015).

CONCLUSION: Placental maternal vascular malperfusion is increased in FCHD. The prevalence of vascular malperfusion did not differ by FCHD type, indicating that CHD type does not predict the likelihood of placental vascular dysfunction. Further investigation of the placental-fetal heart axis in FCHD is warranted given the importance of placental health.

Congenital heart disease affects 1% of infants and is associated with impaired neurodevelopment. Right- or left-sided sulcal features correlate with executive function among people with Tetralogy of Fallot or single ventricle congenital heart disease. Studies of multiple congenital heart disease types are needed to understand regional differences. Further, sulcal pattern has not been studied in people with d-transposition of the great arteries. Therefore, we assessed the relationship between sulcal pattern and executive function, general memory, and processing speed in a meta-regression of 247 participants with three congenital heart disease types (114 single ventricle, 92 d-transposition of the great arteries, and 41 Tetralogy of Fallot) and 94 participants without congenital heart disease. Higher right hemisphere sulcal pattern similarity was associated with improved executive function (Pearson r = 0.19, false discovery rate-adjusted P = 0.005), general memory (r = 0.15, false discovery rate P = 0.02), and processing speed (r = 0.17, false discovery rate P = 0.01) scores. These positive associations remained significant in for the d-transposition of the great arteries and Tetralogy of Fallot cohorts only in multivariable linear regression (estimated change β = 0.7, false discovery rate P = 0.004; β = 4.1, false discovery rate P = 0.03; and β = 5.4, false discovery rate P = 0.003, respectively). Duration of deep hypothermic circulatory arrest was also associated with outcomes in the multivariate model and regression tree analysis. This suggests that sulcal pattern may provide an early biomarker for prediction of later neurocognitive challenges among people with congenital heart disease.

BACKGROUND: Neonatal infections due to Paenibacillus species have increasingly been reported over the last few years.

METHODS: We performed a structured literature review of human Paenibacillus infections in pediatric and adult patients to compare the epidemiology of infections between these distinct patient populations.

RESULTS: Forty reports describing 177 infections were included. Two additional cases were brought to our attention by colleagues. There were 38 Paenibacillus infections occurring in adults caused by 23 species. The clinical presentations of infections were quite variable. In contrast, infections in infants were caused primarily by Paenibacillus thiaminolyticus (112/141, 79%). All the infants with Paenibacillus infection presented with sepsis syndrome or meningitis, often complicated by extensive cerebral destruction and hydrocephalus. Outcomes were commonly poor with 17% (24/141) mortality. Cystic encephalomalacia due to brain destruction was common in both Ugandan and American infant cases and 92/141 (65%) required surgical management of hydrocephalus following their infection.

CONCLUSIONS: Paenibacillus species seem to cause a clinical syndrome in infants characterized by brain abscesses, hydrocephalus and death. This contrasts with infection in adults, which is sporadic with only rare involvement of the central nervous system and very few deaths.

Infant alertness and neurologic changes can reflect life-threatening pathology but are assessed by exam, which can be intermittent and subjective. Reliable, continuous methods are needed. We hypothesized that our computer vision method to track movement, pose AI, could predict neurologic changes in the neonatal intensive care unit (NICU). We collected 4,705 hours of video linked to electroencephalograms (EEG) from 115 infants. We trained a deep learning pose algorithm that accurately predicted anatomic landmarks in three evaluation sets (ROC-AUCs 0.83-0.94), showing feasibility of applying pose AI in an ICU. We then trained classifiers on landmarks from pose AI and observed high performance for sedation (ROC-AUCs 0.87-0.91) and cerebral dysfunction (ROC-AUCs 0.76-0.91), demonstrating that an EEG diagnosis can be predicted from video data alone. Taken together, deep learning with pose AI may offer a scalable, minimally invasive method for neuro-telemetry in the NICU.

Boston Children's Hospital has established a genomic sequencing and analysis research initiative to improve clinical care for pediatric rare disease patients. Through the Children's Rare Disease Collaborative (CRDC), the hospital offers CLIA-grade exome and genome sequencing, along with other sequencing types, to patients enrolled in specialized rare disease research studies. The data, consented for broad research use, are harmonized and analyzed with CRDC-supported variant interpretation tools. Since its launch, 66 investigators representing 26 divisions and 45 phenotype-based cohorts have joined the CRDC. These studies enrolled 4653 families, with 35% of analyzed cases having a finding either confirmed or under further investigation. This accessible and harmonized genomics platform also supports additional institutional data collections, research and clinical, and now encompasses 13,800+ patients and their families. This has fostered new research projects and collaborations, increased genetic diagnoses and accelerated innovative research via integration of genomics research with clinical care.