Critically ill pediatric patients exhibit heterogeneous inflammatory responses associated with differences in mortality risk and treatment response. Prior studies have identified hyper- and hypo-inflammatory subphenotypes using circulating biomarkers and latent class analysis (LCA). However, biomarker assays are costly, time-consuming, and not routinely available in most clinical settings, limiting their use for real-time clinical decision making. This study investigates whether inflammatory subphenotypes can be identified using only routinely collected electronic health record (EHR) variables. Using data from five pediatric critical care cohorts (n = 1,162), we developed an elastic net–penalized logistic regression classifier based on standardized clinical measurements from PRISM-III and PELOD-2 severity scores. The model was trained using only routinely collected clinical variables, while biomarker-derived LCA subphenotypes served as the ground-truth labels. The classifier demonstrated strong discrimination in external validation cohorts (AUROC ≈ 0.90) and identified patients with clinical characteristics consistent with hyper-inflammatory illness. Because the model relies exclusively on widely available clinical variables rather than biomarker assays, it enables rapid, low-cost bedside identification of inflammatory subphenotypes and may support earlier risk stratification and more personalized care in pediatric critical care.

Background: The relationship between triglycerides, white matter hyperintensities (WMH), and regional tau accumulation across the AD spectrum remains poorly characterized. We examined these relationships longitudinally and cross-sectionally across diagnostic groups using [F-18]flortaucipir (FTP) PET and MRI-derived WMH data from the ADNI database. Methods: A consecutive series of subjects included in ADNI through March 2026 were categorized by clinical presentation – cognitively normal (CN), mild cognitive impairment (MCI), or dementia (D) – and were included based on availability of neuroimaging and triglyceride data; for cross-sectional analyses, 2,261 subjects with WMH data contributed multiple timepoint observations (1512 CN, 2,846 MCI, and 468 D, total time-points = 4,826), while 1,357 subjects had FTP PET data (759 CN, 443 MCI, 139 D) available for cross-sectional analyses, among whom those with triglyceride data and at least two FTP scans were included in longitudinal analyses. Tau SUVr values were normalized to inferior cerebellar grey matter and rates of change were calculated as differences per 12 months. WMH volumes were similarly assessed longitudinally, and also examined cross-sectionally. Linear regression analyses for longitudinal data were performed within and across diagnostic groups. Results: Baseline triglycerides showed a significant positive correlation with longitudinal tau accumulation across multiple regions – including the isthmus cingulate (p=0.033), medial orbitofrontal cortex (p=0.031), and amygda

My objective in Dr. Steven Jonas’ laboratory is to see if different lipid compositions can help lipid nanoparticles (LNPs) deliver more CRISPR/Cas9 mRNA and replacement DNA molecules to human bronchial epithelial cells to increase editing efficiency. CRISPR is a precise gene-editing tool that allows scientists to cut and alter DNA at specific locations in living cells. It uses a guide RNA to direct the Cas9 enzyme, to target specific DNA sequences, enabling the deletion, insertion, or modification of genes to correct genetic defects. CF is caused by mutations in the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) gene, making it a perfect candidate for gene-editing technology. CFTR dysfunctions cause normal secretions to pulmonary and other tissues to become covered with thick, sticky mucus, leading to chronic infections and critical damage to the lungs and the pancreas. Our approach delivers genetic cargo through lipid nanoparticles (LNP) that can easily penetrate the outer-cell membrane. Once inserted, the genetic modification can correct the underlying defect which causes CF, taking advantage of the cell membrane’s high fat content. In the lab, LNPs were produced that were confirmed to be carrying the gene editing cargo and were <200nm from cryo-TEM. By adding the cationic lipids DOTAP and MVL5, they helped deliver mRNA and DNA more effectively, allowing for the particles to grow to the desired size and carry the genetic material. The data collected will help decide whether similar relationships exist in LNP-meditated CRISPR/Cas9 delivery to human bronchial epithelial cells. At a broader level, this research addresses two major barriers in gene therapy: overcoming biological barriers to delivery in airway tissue, and improving the efficiency of gene-editing cargoes.

Periprosthetic joint infection (PJI) remains a devastating complication of total joint arthroplasty, yet strain-specific drivers of Staphylococcus aureus osteolysis remain poorly defined. The present study compared four bioluminescent, clinically derived S. aureus strains—three MSSA (Xen29, wound isolate; Xen36, bacteremia isolate; GPF-1, osteomyelitis isolate) and one MRSA (AH4807, community-acquired isolate)—in an established murine post-arthroplasty knee infection model. Longitudinal in vivo bioluminescence imaging tracked bacterial burden over 56 days, while micro-computed tomography quantified bone structural changes at baseline and biweekly intervals. H&E-stained sections assessed histopathology at endpoint. All strains established chronic infection with distinct trajectories. Xen36 and AH4807 showed the highest peak bioluminescence (~1 × 10^6 photons/sec), whereas Xen29 and GPF-1 remained lower (~1 × 10^4 to 1 × 10^5). Despite comparable persistence, micro-computed tomography revealed significant strain-dependent bone loss (p < 0.001): GPF-1 caused the greatest BV/TV reduction (−42.3% vs −5.2% in controls), followed by AH4807 (−28.7%), while Xen36 and Xen29 showed minimal change. GPF-1 also exhibited the greatest trabecular separation and periosteal expansion. These findings demonstrate that strain identity, rather than bacterial burden alone, determines osteolytic severity in PJI, highlighting the need for strain-specific therapeutic strategies to prevent irreversible bone loss beyond conventional antibiotic therapy.