Publications

This paper addresses the problem of detecting possible serious bacterial infection (pSBI) of infancy, i.e. a clinical presentation consistent with bacterial sepsis in newborn infants using cranial ultrasound (cUS) images. The captured image set for each patient enables multiview imagery: coronal and sagittal, with geometric overlap. To exploit this geometric relation, we develop a new learning framework, called the intersection-guided Crossview Local- and Image-level Fusion Network (CLIF-Net). Our technique employs two distinct convolutional neural network branches to extract features from coronal and sagittal images with newly developed multi-level fusion blocks. Specifically, we leverage the spatial position of these images to locate the intersecting region. We then identify and enhance the semantic features from this region across multiple levels using cross-attention modules, facilitating the acquisition of mutually beneficial and more representative features from both views. The final enhanced features from the two views are then integrated and projected through the image-level fusion layer, outputting pSBI and non-pSBI class probabilities. We contend that our method of exploiting multi-view cUS images enables a first of its kind, robust 3D representation tailored for pSBI detection. When evaluated on a dataset of 302 cUS scans from Mbale Regional Referral Hospital in Uganda, CLIF-Net demonstrates substantially enhanced performance, surpassing the prevailing state-of-the-art infection detection techniques.

OBJECTIVE: To increase length board use for eligible neonatal intensive care unit (NICU) infants.

STUDY DESIGN: We implemented a quality improvement study involving 704 infants in our level IV NICU. A multidisciplinary workgroup developed guidelines for length measurement technique and completed Plan-Do-Study-Act cycles. Outcome measure was the weekly proportion of eligible infants who received a length board measurement. Process measures were the weekly proportion of infants with any length measurement or had method documented. Balancing measure was the incidence of unplanned dislodgements of drains, tubes, or catheters.

RESULTS: After the guideline launch, both process measure proportions increased. Weekly mean percentage of length board increased from 11 to 63%. There were no dislodgement events. Length board measurements were less likely to demonstrate a negative change week-to-week (10% vs 18%, 16/161 vs 59/326, Fisher p = 0.02).

CONCLUSION: In a level IV NICU, a quality improvement initiative increased the safe use of length boards.

The use of genomic sequencing (GS) for prenatal diagnosis of fetuses with sonographic abnormalities has grown tremendously over the past decade. Fetal GS also offers an opportunity to identify incidental genomic variants that are unrelated to the fetal phenotype but may be relevant to fetal and newborn health. There are currently no guidelines for reporting incidental findings from fetal GS. In the United States, GS for adults and children is recommended to include a list of "secondary findings" genes (ACMG SF v.3.2) that are associated with disorders for which surveillance or treatment can reduce morbidity and mortality. The genes on ACMG SF v.3.2 predominantly cause adult-onset disorders. Importantly, many genetic disorders with fetal and infantile onset are treatable as well. A proposed solution is to create a "treatable fetal findings list," which can be offered to pregnant individuals undergoing fetal GS or, eventually, as a standalone cell-free fetal DNA screening test. In this integrative review, we propose criteria for a treatable fetal findings list, then identify genetic disorders with clinically available or emerging fetal interventions and those for which clinical detection and intervention in the first week of life might lead to improved outcomes. Finally, we synthesize the potential benefits, limitations, and risks of a treatable fetal findings list.

Functional impact of noncoding variants can be predicted using computational approaches. Although predictive scores can be insightful, implementing the scores for a custom variant set and associating scores with complex traits require multiple phases of analysis. Here, we present a protocol for prioritizing variants by generating deep-learning-predicted functional scores and relating them with brain traits. We describe steps for score prediction, statistical comparison, phenotype correlation, and functional enrichment analysis. This protocol can be generalized to different models and phenotypes. For complete details on the use and execution of this protocol, please refer to Mondragon-Estrada et al.1.

Congenital heart disease (CHD) is a leading cause of infant mortality. We analyzed de novo mutations (DNMs) and very rare transmitted/unphased damaging variants in 248 prespecified genes in 11,555 CHD probands. The results identified 60 genes with a significant burden of heterozygous damaging variants. Variants in these genes accounted for CHD in 10.1% of probands with similar contributions from de novo and transmitted variants in parent-offspring trios that showed incomplete penetrance. DNMs in these genes accounted for 58% of the signal from DNMs. Thirty-three genes were linked to a single CHD subtype while 12 genes were associated with 2 to 4 subtypes. Seven genes were only associated with isolated CHD, while 37 were associated with 1 or more extracardiac abnormalities. Genes selectively expressed in the cardiomyocyte lineage were associated with isolated CHD, while those widely expressed in the brain were also associated with neurodevelopmental delay (NDD). Missense variants introducing or removing cysteines in epidermal growth factor (EGF)-like domains of NOTCH1 were enriched in tetralogy of Fallot and conotruncal defects, unlike the broader CHD spectrum seen with loss of function variants. Transmitted damaging missense variants in MYH6 were enriched in multiple CHD phenotypes and account for  1% of all probands. Probands with characteristic mutations causing syndromic CHD were frequently not diagnosed clinically, often due to missing cardinal phenotypes. CHD genes that were positively or negatively associated with development of NDD suggest clinical value of genetic testing. These findings expand the understanding of CHD genetics and support the use of molecular diagnostics in CHD.

Examining the altered arrangement and patterning of sulcal folds offers insights into the mechanisms of neurodevelopmental differences in psychiatric and neurological disorders. Previous sulcal pattern analysis used spectral graph matching of sulcal pit-based graph structures to assess deviations from normative sulcal patterns. However, challenges exist, including the absence of a standard criterion for defining a typical reference set, time-consuming cost of graph matching, user-defined feature weight sets, and assumptions about uniform node distribution. We developed a deep learning-based sulcal pattern analysis to address these challenges by adapting prototype-based graph neural networks to sulcal pattern graphs. Additionally, we proposed a prototype inverse-projection for better interpretability. Unlike other prototype-based models, our approach inversely projects prototypes onto individual node representations to calculate the inverse-projection weights, enabling efficient visualization of prototypes and focusing the model on selective regions. We evaluated our method through a classification task between healthy controls (n = 174, age = 15.4 ±1.9 [mean ± standard deviation, years]) and patients with congenital heart disease (n = 345, age = 15.8 ±4.7) from four cohort studies and a public dataset. Our approach demonstrated superior classification performance compared to other state-of-the-art models, supported by extensive ablative studies. Furthermore, we visualized and examined the learned prototypes to enhance understanding. We believe our method has the potential to be a sensitive and understandable tool for sulcal pattern analysis.

Variants with large effect contribute to congenital heart disease (CHD). To date, recessive genotypes (RGs) have commonly been implicated through anecdotal ascertainment of consanguineous families and candidate gene-based analysis; the recessive contribution to the broad range of CHD phenotypes has been limited. We analyzed whole exome sequences of 5,424 CHD probands. Rare damaging RGs were estimated to contribute to at least 2.2% of CHD, with greater enrichment among laterality phenotypes (5.4%) versus other subsets (1.4%). Among 108 curated human recessive CHD genes, there were 66 RGs, with 54 in 11 genes with >1 RG, 12 genes with 1 RG, and 85 genes with zero. RGs were more prevalent among offspring of consanguineous union (4.7%, 32/675) than among nonconsanguineous probands (0.7%, 34/4749). Founder variants in GDF1 and PLD1 accounted for 74% of the contribution of RGs among 410 Ashkenazi Jewish probands. We identified genome-wide significant enrichment of RGs in C1orf127, encoding a likely secreted protein expressed in embryonic mouse notochord and associated with laterality defects. Single-cell transcriptomes from gastrulation-stage mouse embryos revealed enrichment of RGs in genes highly expressed in the cardiomyocyte lineage, including contractility-related genes MYH6, UNC45B, MYO18B, and MYBPC3 in probands with left-sided CHD, consistent with abnormal contractile function contributing to these malformations. Genes with significant RG burden account for 1.3% of probands, more than half the inferred total. These results reveal the recessive contribution to CHD, and indicate that many genes remain to be discovered, with each likely accounting for a very small fraction of the total.

BACKGROUND: SMAD2 is a coregulator that binds a variety of transcription factors in human development. Heterozygous SMAD2 loss-of-function and missense variants are identified in patients with congenital heart disease (CHD) or arterial aneurysms. Mechanisms that cause distinct cardiovascular phenotypes remain unknown. We aimed to define transcriptional and epigenetic effects of SMAD2 variants and their role in CHD. We also assessed the function of SMAD2 missense variants of uncertain significance.

METHODS AND RESULTS: Rare SMAD2 variants (minor allele frequency ≤10-5) were identified in exome sequencing of 11 336 participants with CHD. We constructed isogenic induced pluripotent stem cells with heterozygous or homozygous loss-of-function and missense SMAD2 variants identified in CHD probands. Wild-type and mutant induced pluripotent stem cells were analyzed using bulk RNA sequencing, chromatin accessibility (Assay for Transposase-Accessible Chromatin With Sequencing), and integrated with published SMAD2/3 chromatin immunoprecipitation data. Cardiomyocyte differentiation and contractility were evaluated. Thirty participants with CHD had heterozygous loss-of-function or missense SMAD2 variants. SMAD2 haploinsufficiency altered chromatin accessibility at promoters and dysregulated expression of 385 SMAD regulated genes, including 10 CHD-associated genes. Motifs enriched in differential Assay for Transposase-Accessible Chromatin peaks predicted that SMAD2 haploinsufficiency disrupts interactions with transcription factors NANOG (homeobox protein NANOG), ETS, TEAD3/4 (transcriptional enhanced associate domain 3/4), CREB1 (cAMP response element binding protein 1), and AP1 (activator protein 1). Compared with SMAD2-haploinsufficient cells, induced pluripotent stem cells with R114C or W274C variants exhibited distinct and shared chromatin accessibility and transcription factor binding changes.

CONCLUSIONS: SMAD2 haploinsufficiency disrupts transcription factor binding and chromatin interactions critical for cardiovascular development. Differences between the molecular consequences of loss-of-function and missense variants likely contribute to phenotypic heterogeneity. These findings indicate opportunities for molecular analyses to improve reclassification of SMAD2 variants of uncertain clinical significance.

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.