Detecting pediatric wrist fractures using deep-learning-based object detection.

BACKGROUND: Missed fractures are the leading cause of diagnostic error in the emergency department, and fractures of pediatric bones, particularly subtle wrist fractures, can be misidentified because of their varying characteristics and responses to injury. OBJECTIVE: This study evaluated the utility of an object detection deep learning framework for classifying pediatric wrist fractures as positive or negative for fracture, including subtle buckle fractures of the distal radius, and evaluated the performance of this algorithm as augmentation to trainee radiograph interpretation. MATERIALS AND METHODS: We obtained 395 posteroanterior wrist radiographs from unique pediatric patients (65% positive for fracture, 30% positive for distal radial buckle fracture) and divided them into train (n = 229), tune (n = 41) and test (n = 125) sets. We trained a Faster R-CNN (region-based convolutional neural network) deep learning object-detection model. Two pediatric and two radiology residents evaluated radiographs initially without the artificial intelligence (AI) assistance, and then subsequently with access to the bounding box generated by the Faster R-CNN model. RESULTS: The Faster R-CNN model demonstrated an area under the curve (AUC) of 0.92 (95% confidence interval [CI] 0.87-0.97), accuracy of 88% (n = 110/125; 95% CI 81-93%), sensitivity of 88% (n = 70/80; 95% CI 78-94%) and specificity of 89% (n = 40/45, 95% CI 76-96%) in identifying any fracture and identified 90% of buckle fractures (n = 35/39, 95% CI 76-97%). Access to Faster R-CNN model predictions significantly improved average resident accuracy from 80 to 93% in detecting any fracture (P < 0.001) and from 69 to 92% in detecting buckle fracture (P < 0.001). After accessing AI predictions, residents significantly outperformed AI in cases of disagreement (73% resident correct vs. 27% AI, P = 0.002). CONCLUSION: An object-detection-based deep learning approach trained with only a few hundred examples identified radiographs containing pediatric wrist fractures with high accuracy. Access to model predictions significantly improved resident accuracy in diagnosing these fractures.

A 12-Year-Old Girl with Weight Loss.

A 12-Year-Old Girl with Weight LossThis report examines a case of chronic weight loss, anxiety, abdominal pain, and nausea in an adolescent girl. With directed questioning, physical examination, and testing, an illness script for the presentation emerges; the differential is refined until a final diagnosis is made.

Canakinumab for the treatment of autoinflammatory very early onset- inflammatory bowel disease.

Introduction: Therapeutic options are critically needed for children with refractory very early onset inflammatory bowel disease (VEO-IBD). Our aim was to evaluate clinical response to canakinumab, an anti-IL-1ß monoclonal antibody, in patients with VEO-IBD whose phenotype resembles those with monogenic autoinflammatory disease. Methods: This is a single center retrospective study of patients with VEO-IBD with autoinflammatory phenotype (AIP) in the absence of identified monogenic disease treated with canakinumab for >6 months. AIP was defined as confirmed IBD with associated signs of systemic inflammation in the absence of infection, including leukocytosis, markedly elevated inflammatory markers, and extraintestinal manifestations (recurrent fevers, oral ulcers, arthritis). Primary outcomes included clinical response in disease activity indices after 6 months of therapy. Secondary outcomes included rate of AIP signs and symptoms, growth, surgery, steroid use, hospitalizations, and adverse events. Results: Nineteen patients were included: 47% with infantile onset, 58% classified as IBD-U, and 42% classified as CD. At baseline, 37% were biologic naïve, and canakinumab was used as dual therapy in 74% of patients. Clinical response was achieved in 89% with statistically significant improvement in PCDAI and PUCAI. Clinical remission was achieved in 32% of patients. There was significant improvement in the clinical manifestations of AIP and the biochemical markers of disease. Number of hospitalizations (p<0.01) and length of stay (p<0.05) decreased. Growth improved with median weight-for-length Z-score increasing from -1.01 to 1.1 in children less than 2 years old. There were minimal adverse events identified during the study period. Conclusion: Canakinumab may be an effective and safe treatment for a subset of children with VEO-IBD with AIP, as well as older patients with IBD. This study highlights the importance of a precision medicine approach in children with VEO-IBD.

An immunogenomic phenotype predicting behavioral treatment response: Toward precision psychiatry for mothers and children with trauma exposure.

Inflammatory pathways predict antidepressant treatment non-response among individuals with major depression; yet, this phenomenon may have broader transdiagnostic and transtherapeutic relevance. Among trauma-exposed mothers (Mage = 32 years) and their young children (Mage = 4 years), we tested whether genomic and proteomic biomarkers of pro-inflammatory imbalance prospectively predicted treatment response (PTSD and depression) to an empirically-supported behavioral treatment. Forty-three mother-child dyads without chronic disease completed Child Parent Psychotherapy (CPP) for roughly 9 months. Maternal blood was drawn pre-treatment, CD14 + monocytes isolated, gene expression derived from RNA sequencing (n = 34; Illumina HiSeq 4000;TruSeqcDNA library), and serum assayed (n = 43) for C-Reactive Protein (CRP) and interleukin-1ß (IL-1ß). Symptoms of PTSD and depression decreased significantly from pre- to post-treatment for both mothers and children (all p's < 0.01). Nonetheless, a higher pre-treatment maternal pro-inflammatory imbalance of M1-like versus M2-like macrophage-associated RNA expression (M1/M2) (ß = 0.476, p = .004) and IL-1ß (ß=0.333, p = .029), but not CRP, predicted lesser improvements in maternal PTSD symptoms, unadjusted and adjusting for maternal age, BMI, ethnicity, antidepressant use, income, education, and US birth. Only higher pre-treatment M1/M2 predicted a clinically-relevant threshold of PTSD non-response among mothers (OR = 3.364, p = .015; ROC-AUC = 0.78). Additionally, higher M1/M2 predicted lesser decline in maternal depressive symptoms (ß = 0.556, p = .001), though not independent of PTSD symptoms. For child outcomes, higher maternal IL-1ß significantly predicted poorer PTSD and depression symptom trajectories (ß's = 0.318-0.429, p's < 0.01), while M1/M2 and CRP were marginally associated with poorer PTSD symptom improvement (ß's = 0.295-0.333, p's < 0.056). Pre-treatment pro-inflammatory imbalance prospectively predicts poorer transdiagnostic symptom response to an empirically-supported behavioral treatment for trauma-exposed women and their young children.

Adversity in early life and pregnancy are immunologically distinct from total life adversity: macrophage-associated phenotypes in women exposed to interpersonal violence.

Early childhood and pregnancy are two sensitive periods of heightened immune plasticity, when exposure to adversity may disproportionately increase health risks. However, we need deeper phenotyping to disentangle the impact of adversity during sensitive periods from that across the total lifespan. This study examined whether retrospective reports of adversity during childhood or pregnancy were associated with inflammatory imbalance, in an ethnically diverse cohort of 53 low-income women seeking family-based trauma treatment following exposure to interpersonal violence. Structured interviews assessed early life adversity (trauma exposure ≤ age 5), pregnancy adversity, and total lifetime adversity. Blood serum was assayed for pro-inflammatory (TNF-a, IL-1ß, IL-6, and CRP) and anti-inflammatory (IL-1RA, IL-4, and IL-10) cytokines. CD14+ monocytes were isolated in a subsample (n = 42) and gene expression assayed by RNA sequencing (Illumina HiSeq 4000; TruSeq cDNA library). The primary outcome was a macrophage-associated M1/M2 gene expression phenotype. To evaluate sensitivity and specificity, we contrasted M1/M2 gene expression with a second, clinically-validated macrophage-associated immunosuppressive phenotype (endotoxin tolerance) and with pro-inflammatory and anti-inflammatory cytokine levels. Adjusting for demographics, socioeconomic status, and psychopathology, higher adversity in early life (ß = .337, p = 0.029) and pregnancy (ß = .332, p = 0.032) were each associated with higher M1/M2 gene expression, whereas higher lifetime adversity (ß = -.341, p = 0.031) was associated with lower immunosuppressive gene expression. Adversity during sensitive periods was uniquely associated with M1/M2 imbalance, among low-income women with interpersonal violence exposure. Given that M1/M2 imbalance is found in sepsis, severe COVID-19 and myriad chronic diseases, these findings implicate novel immune mechanisms underlying the impact of adversity on health.

FCRL5+ Memory B Cells Exhibit Robust Recall Responses.

FCRL5+ atypical memory B cells (atMBCs) expand in many chronic human infections, including recurrent malaria, but studies have drawn opposing conclusions about their function. Here, in mice infected with Plasmodium chabaudi, we demonstrate expansion of an antigen-specific FCRL5+ population that is distinct from previously described FCRL5+ innate-like murine subsets. Comparative analyses reveal overlapping phenotypic and transcriptomic signatures between FCRL5+ B cells from Plasmodium-infected mice and atMBCs from Plasmodium-exposed humans. In infected mice, FCRL5 is expressed on the majority of antigen-specific germinal-center-derived memory B cells (MBCs). Upon challenge, FCRL5+ MBCs rapidly give rise to antibody-producing cells expressing additional atypical markers, demonstrating functionality in vivo. Moreover, atypical markers are expressed on antigen-specific MBCs generated by immunization in both mice and humans, indicating that the atypical phenotype is not restricted to chronic settings. This study resolves conflicting interpretations of atMBC function and suggests FCRL5+ B cells as an attractive target for vaccine strategies.

Empirical assessment of the impact of sample number and read depth on RNA-Seq analysis workflow performance.

BACKGROUND: RNA-Sequencing analysis methods are rapidly evolving, and the tool choice for each step of one common workflow, differential expression analysis, which includes read alignment, expression modeling, and differentially expressed gene identification, has a dramatic impact on performance characteristics. Although a number of workflows are emerging as high performers that are robust to diverse input types, the relative performance characteristics of these workflows when either read depth or sample number is limited-a common occurrence in real-world practice-remain unexplored. RESULTS: Here, we evaluate the impact of varying read depth and sample number on the performance of differential gene expression identification workflows, as measured by precision, or the fraction of genes correctly identified as differentially expressed, and by recall, or the fraction of differentially expressed genes identified. We focus our analysis on 30 high-performing workflows, systematically varying the read depth and number of biological replicates of patient monocyte samples provided as input. We find that, in general for most workflows, read depth has little effect on workflow performance when held above two million reads per sample, with reduced workflow performance below this threshold. The greatest impact of decreased sample number is seen below seven samples per group, when more heterogeneity in workflow performance is observed. The choice of differential expression identification tool, in particular, has a large impact on the response to limited inputs. CONCLUSIONS: Among the tested workflows, the recall/precision balance remains relatively stable at a range of read depths and sample numbers, although some workflows are more sensitive to input restriction. At ranges typically recommended for biological studies, performance is more greatly impacted by the number of biological replicates than by read depth. Caution should be used when selecting analysis workflows and interpreting results from low sample number experiments, as all workflows exhibit poorer performance at lower sample numbers near typically reported values, with variable impact on recall versus precision. These analyses highlight the performance characteristics of common differential gene expression workflows at varying read depths and sample numbers, and provide empirical guidance in experimental and analytical design.

Empirical assessment of analysis workflows for differential expression analysis of human samples using RNA-Seq.

BACKGROUND: RNA-Seq has supplanted microarrays as the preferred method of transcriptome-wide identification of differentially expressed genes. However, RNA-Seq analysis is still rapidly evolving, with a large number of tools available for each of the three major processing steps: read alignment, expression modeling, and identification of differentially expressed genes. Although some studies have benchmarked these tools against gold standard gene expression sets, few have evaluated their performance in concert with one another. Additionally, there is a general lack of testing of such tools on real-world, physiologically relevant datasets, which often possess qualities not reflected in tightly controlled reference RNA samples or synthetic datasets. RESULTS: Here, we evaluate 219 combinatorial implementations of the most commonly used analysis tools for their impact on differential gene expression analysis by RNA-Seq. A test dataset was generated using highly purified human classical and nonclassical monocyte subsets from a clinical cohort, allowing us to evaluate the performance of 495 unique workflows, when accounting for differences in expression units and gene- versus transcript-level estimation. We find that the choice of methodologies leads to wide variation in the number of genes called significant, as well as in performance as gauged by precision and recall, calculated by comparing our RNA-Seq results to those from four previously published microarray and BeadChip analyses of the same cell populations. The method of differential gene expression identification exhibited the strongest impact on performance, with smaller impacts from the choice of read aligner and expression modeler. Many workflows were found to exhibit similar overall performance, but with differences in their calibration, with some biased toward higher precision and others toward higher recall. CONCLUSIONS: There is significant heterogeneity in the performance of RNA-Seq workflows to identify differentially expressed genes. Among the higher performing workflows, different workflows exhibit a precision/recall tradeoff, and the ultimate choice of workflow should take into consideration how the results will be used in subsequent applications. Our analyses highlight the performance characteristics of these workflows, and the data generated in this study could also serve as a useful resource for future development of software for RNA-Seq analysis.

A Novel Model of Asymptomatic Plasmodium Parasitemia That Recapitulates Elements of the Human Immune Response to Chronic Infection.

In humans, immunity to Plasmodium sp. generally takes the form of protection from symptomatic malaria (i.e., 'clinical immunity') rather than infection ('sterilizing immunity'). In contrast, mice infected with Plasmodium develop sterilizing immunity, hindering progress in understanding the mechanistic basis of clinical immunity. Here we present a novel model in which mice persistently infected with P. chabaudi exhibit limited clinical symptoms despite sustaining patent parasite burdens for many months. Characterization of immune responses in persistently infected mice revealed development of CD4+ T cell exhaustion, increased production of IL-10, and expansion of B cells with an atypical surface phenotype. Additionally, persistently infected mice displayed a dramatic increase in circulating nonclassical monocytes, a phenomenon that we also observed in humans with both chronic Plasmodium exposure and asymptomatic infection. Following pharmacological clearance of infection, previously persistently infected mice could not control a secondary challenge, indicating that persistent infection disrupts the sterilizing immunity that typically develops in mouse models of acute infection. This study establishes an animal model of asymptomatic, persistent Plasmodium infection that recapitulates several central aspects of the immune response in chronically exposed humans. As such, it provides a novel tool for dissection of immune responses that may prevent development of sterilizing immunity and limit pathology during infection.

Trimming of sequence reads alters RNA-Seq gene expression estimates.

BACKGROUND: High-throughput RNA-Sequencing (RNA-Seq) has become the preferred technique for studying gene expression differences between biological samples and for discovering novel isoforms, though the techniques to analyze the resulting data are still immature. One pre-processing step that is widely but heterogeneously applied is trimming, in which low quality bases, identified by the probability that they are called incorrectly, are removed. However, the impact of trimming on subsequent alignment to a genome could influence downstream analyses including gene expression estimation; we hypothesized that this might occur in an inconsistent manner across different genes, resulting in differential bias. RESULTS: To assess the effects of trimming on gene expression, we generated RNA-Seq data sets from four samples of larval Drosophila melanogaster sensory neurons, and used three trimming algorithms--SolexaQA, Trimmomatic, and ConDeTri-to perform quality-based trimming across a wide range of stringencies. After aligning the reads to the D. melanogaster genome with TopHat2, we used Cuffdiff2 to compare the original, untrimmed gene expression estimates to those following trimming. With the most aggressive trimming parameters, over ten percent of genes had significant changes in their estimated expression levels. This trend was seen with two additional RNA-Seq data sets and with alternative differential expression analysis pipelines. We found that the majority of the expression changes could be mitigated by imposing a minimum length filter following trimming, suggesting that the differential gene expression was primarily being driven by spurious mapping of short reads. Slight differences with the untrimmed data set remained after length filtering, which were associated with genes with low exon numbers and high GC content. Finally, an analysis of paired RNA-seq/microarray data sets suggests that no or modest trimming results in the most biologically accurate gene expression estimates. CONCLUSIONS: We find that aggressive quality-based trimming has a large impact on the apparent makeup of RNA-Seq-based gene expression estimates, and that short reads can have a particularly strong impact. We conclude that implementation of trimming in RNA-Seq analysis workflows warrants caution, and if used, should be used in conjunction with a minimum read length filter to minimize the introduction of unpredictable changes in expression estimates.

JUNB is a key transcriptional modulator of macrophage activation.

Activated macrophages are crucial for restriction of microbial infection but may also promote inflammatory pathology in a wide range of both infectious and sterile conditions. The pathways that regulate macrophage activation are therefore of great interest. Recent studies in silico have putatively identified key transcription factors that may control macrophage activation, but experimental validation is lacking. In this study, we generated a macrophage regulatory network from publicly available microarray data, employing steps to enrich for physiologically relevant interactions. Our analysis predicted a novel relationship between the AP-1 family transcription factor Junb and the gene Il1b, encoding the pyrogen IL-1ß, which macrophages express upon activation by inflammatory stimuli. Previously, Junb has been characterized primarily as a negative regulator of the cell cycle, whereas AP-1 activity in myeloid inflammatory responses has largely been attributed to c-Jun. We confirmed experimentally that Junb is required for full expression of Il1b, and of additional genes involved in classical inflammation, in macrophages treated with LPS and other immunostimulatory molecules. Furthermore, Junb modulates expression of canonical markers of alternative activation in macrophages treated with IL-4. Our results demonstrate that JUNB is a significant modulator of both classical and alternative macrophage activation. Further, this finding provides experimental validation for our network modeling approach, which will facilitate the future use of gene expression data from open databases to reveal novel, physiologically relevant regulatory relationships.

Loss of Toll-like receptor 7 alters cytokine production and protects against experimental cerebral malaria.

BACKGROUND: Malaria, caused by Plasmodium sp. parasites, is a leading cause of global morbidity and mortality. Cerebral malaria, characterized by neurological symptoms, is a life-threatening complication of malaria affecting over 500,000 young children in Africa every year. Because of the prevalence and severity of cerebral malaria, a better understanding of the underlying molecular mechanisms of its pathology is desirable and could inform future development of therapeutics. This study sought to clarify the role of Toll-like receptors (TLRs) in promoting immunopathology associated with cerebral malaria, with a particular focus on the understudied TLR7. METHODS: Using the Plasmodium berghei ANKA mouse model of experimental cerebral malaria, C57BL/6 mice deficient in various TLRs were infected, and their resistance to cerebral malaria and immune activation through cytokine production were measured. RESULTS: Loss of TLR7 conferred partial protection against fatal experimental cerebral malaria. Additionally, loss of TLR signalling dysregulated the cytokine profile, resulting in a shift in the cytokine balance towards those with more anti-inflammatory properties. CONCLUSION: This work identifies signalling through TLR7 as a source of pathology in experimental cerebral malaria.

Toll-like receptor 7 mediates early innate immune responses to malaria.

Innate immune recognition of malaria parasites is the critical first step in the development of the host response. At present, Toll-like receptor 9 (TLR9) is thought to play a central role in sensing malaria infection. However, we and others have observed that Tlr9(-/-) mice, in contrast to mice deficient in the downstream adaptor, Myeloid differentiation primary response gene 88 (MYD88), exhibit few deficiencies in immune function during early infection with the malaria parasite Plasmodium chabaudi, implying that another MYD88-dependent receptor also contributes to the antimalarial response. Here we use candidate-based screening to identify TLR7 as a key sensor of early P. chabaudi infection. We show that TLR7 mediates a rapid systemic response to infection through induction of cytokines such as type I interferons (IFN-I), interleukin 12, and gamma interferon. TLR7 is also required for induction of IFN-I by other species and strains of Plasmodium, including an etiological agent of human disease, P. falciparum, suggesting that malaria parasites harbor a common pathogen-associated molecular pattern (PAMP) recognized by TLR7. In contrast to the nonredundant requirement for TLR7 in early immune activation, sensing through both TLR7 and TLR9 was required for proinflammatory cytokine production and immune cell activation during the peak of parasitemia. Our findings indicate that TLR7 plays a central role in early immune activation during malaria infection, whereas TLR7 and TLR9 contribute combinatorially to immune responses as infection progresses.