Publications by Year: 2026

IMPORTANCE: Bronchopulmonary dysplasia (BPD) and pulmonary hypertension (PH) are leading causes of morbidity and mortality in premature infants.

OBJECTIVE: To determine whether images obtained as part of retinopathy of prematurity (ROP) screening might contain features associated with BPD and PH in infants and whether a multimodal model integrating imaging features with demographic risk factors might outperform a model based on demographic risk alone.

DESIGN, SETTING, AND PARTICIPANTS: A deep learning model was used to study retinal images collected from patients enrolled in the multi-institutional Imaging and Informatics in Retinopathy of Prematurity (i-ROP) study. The analysis included infants at risk for ROP undergoing routine ROP screening examinations from 2012 to 2020. Infants were recruited from 7 neonatal intensive care units. Images were limited to 34 weeks' or less postmenstrual age (PMA) so as to precede the clinical diagnosis of BPD or PH. The dataset included the period from June 2015 to April 2020. Data were analyzed from April to June 2025.

EXPOSURES: BPD was diagnosed by the presence of an oxygen requirement at 36 weeks' PMA, and PH was diagnosed by echocardiogram at 34 weeks. A support vector machine model was trained to predict BPD or PH diagnosis using (1) image features alone (extracted using ResNet18), (2) demographics alone, or (3) image features concatenated with demographics. To reduce the possibility of confounding with ROP, secondary models were trained using only images without clinical signs of ROP.

MAIN OUTCOMES AND MEASURES: For both BPD and PH, performance was reported on a held-out test set and assessed by the area under receiver operating characteristic curve (AUROC).

RESULTS: A total of 493 infants (mean [SD] gestational age, BPD, 25.7 [1.8] weeks; normal, 27.3 [1.8] weeks; 267 male [54.2%]) were included in this analysis. Performance was reported on a held-out test set (99 patients from the BPD cohort and 37 patients from the PH cohort). For BPD, the multimodal model showed higher accuracy (AUC, 0.82; 95% CI, 0.72-0.90) than demographics-only (0.72; ∆AUC, 0.1; 95% CI, -0.008 to 0.21; P = .07) or imaging-only (0.72; ∆AUC, 0.1; 95% CI, 0.04-0.16; P = .002) models. For PH, multimodal AUC was 0.91 vs the demographics-only 0.68 (∆AUC, 0.14; 95% CI, 0.006-0.27; P = .04) and imaging-only 0.91 (∆AUC, -0.09; 95% CI, -0.3 to 0.12; P = .40) models. Results persisted when trained on images lacking clinical ROP signs.

CONCLUSIONS AND RELEVANCE: Results suggest that retinal images obtained during ROP screening may be used to predict the diagnosis of BPD and PH in preterm infants, which may lead to earlier diagnosis and avoid the need for invasive diagnostic testing in the future.

BACKGROUND: The 2022 American Heart Association/American College of Cardiology/Heart Failure Society of America heart failure (HF) classification incorporates cardiac biomarkers to identify early risk of HF. The HF stages may also guide the prognosis and management of cardiovascular and kidney-related events.

METHODS: SCORED (Effect of Sotagliflozin on Cardiovascular and Renal Events in Patients With Type 2 Diabetes and Moderate Renal Impairment Who Are at Cardiovascular Risk) was a randomized trial in diabetes with kidney disease comparing sotagliflozin versus placebo on cardiovascular death, HF hospitalizations, and urgent HF visits. SCORED participants were grouped by HF stage post hoc. Stage A: no HF, normal biomarkers (NT-proBNP [N-terminal pro-B-type natriuretic peptide] <125 pg/mL; high sensitivity cardiac troponin T [hs-cTnT] <14 ng/L), and normal cardiac structure/function. Stage B (pre-HF): no HF but elevated NT-proBNP, hs-cTnT, or abnormal cardiac structure/function. Stage C/D: symptomatic HF. End points include the primary composite (cardiovascular death and HF-related events), major adverse cardiovascular events, and kidney-related composites (≥50% decline in estimated glomerular filtration rate, kidney failure, or kidney death). Using competing-risk proportional hazards models, we examined the association between HF stage and these end points, and the effect of sotagliflozin versus placebo by HF stage.

RESULTS: There were 741 patients (7%) in stage A, 6560 (62%) in stage B (pre-HF), and 3283 (31%) in stage C/D (established HF). The median NT-proBNP and hs-cTnT increased with HF stage. Increasing HF stage was associated with a 2- to 4-fold increase in the primary outcome/major adverse cardiovascular events in the placebo group. The kidney-specific composite was 5-fold higher in stage B (pre-HF) versus stage A but similar in stages B and C/D. The effect of sotagliflozin versus placebo was similar, irrespective of HF stage (primary outcome: hazard ratio, 0.74 [95% CI, 0.63-0.88]; Pinteraction=1.00), with higher absolute benefit in each HF stage (P-trendIRR=0.002). The absolute benefit for the kidney-specific end point was comparable for stages B and C/D.

CONCLUSIONS: Increasing HF stage is associated with a higher risk of HF, major adverse cardiovascular events, and kidney events. Asymptomatic stage B (pre-HF) increased cardiovascular and renal events by >2- and 5-fold, respectively. The benefits of sotagliflozin are consistent, irrespective of HF stage.

AIMS/HYPOTHESIS: DNA methylation has been shown to be associated with kidney function and diabetic kidney disease (DKD), but prospective studies are scarce. Therefore, we conducted epigenome-wide association studies (EWASs) on early- and late-stage DKD progression using DNA methylation data obtained by analysing baseline blood samples from participants in the Finnish Diabetic Nephropathy Study type 1 diabetes cohort.

METHODS: We included 403 individuals with normal AER (early-stage progression group) and 372 individuals with severe albuminuria (late-stage progression group), and followed up DKD progression, defined as a decrease in eGFR to <60 ml/min per 1.73 m2 in the early-stage progression group, and end-stage kidney disease (ESKD) in the late-stage group. Replication was conducted in two type 1 diabetes cohorts in addition to publicly available EWAS summary statistics from diabetes and general population cohorts. Significant loci were further characterised by integration with genetic and proteomic data.

RESULTS: We identified 11 methylation sites associated with DKD progression (p<9.4 × 10-8). Methylation at cg01730944 near the podocyte-specific gene CDKN1C and three other CpGs associated with early-stage DKD progression were independent of baseline eGFR, whereas late-stage progression CpGs were strongly associated with eGFR. The identified lead ESKD risk locus cg17944885 (chr19p13.2, p=2.6 × 10-17) and several novel methylation sites associated with late-stage DKD progression were supported by the results of previous studies. Proteomic analysis of cis proteins identified potential target genes for two CpGs: cg14999724 methylation was associated with PRG3 and PRG2, and cg12272104 was associated with BSG, FSTL3 and PALM. Furthermore, UK Biobank data show associations between these proteins and severe kidney endpoints. Finally, survival models that included methylation markers in addition to clinical risk factors significantly improved the identification of individuals at risk of early-stage DKD progression.

CONCLUSIONS/INTERPRETATION: The current study detected 11 loci associated with DKD progression, identifying methylation changes predictive of early-stage DKD progression in type 1 diabetes for the first time. Future research is needed to establish prognostic DNA methylation markers for DKD progression.

BACKGROUND: Intracranial mesenchymal tumors, FET::CREB fusion-positive (ICMT), show fusions involving FET RNA-binding protein family genes (EWSR1 or FUS) and CREB family of transcription factors (ATF1, CREB1 or CREM). The methylation signature(s), gene expression characteristics and clinical behavior of this important tumor type require further characterization.

METHODS: We study the methylation profiles of 81 ICMT cases (61 newly profiled cases and 20 cases from publicly available sources). Clinicopathologic and genomic data were recorded for each case when available.

RESULTS: ICMT showed a relatively distinct methylation signature compared to related tumors. Among the 65 cases where fusion types were documented, the identified fusions included EWSR1::ATF1 (25 cases), EWSR1::CREB1 (12 cases), EWSR1::CREM (21 cases), FUS::CREM (3 cases) and SMARCA2::CREM (4 cases). We confirmed the prior description of two distinct subgroups of ICMT (subclasses A and B). The majority of the cases belonged to subclass A (n = 69; 85%), which showed higher median age compared to subclass B patients (26 years vs. 15 years). Subclass B cases (n = 12; 15%) showed shorter progression-free survival (p < 0.01). Gene expression analysis of ICMT showed key overexpressed markers in ICMT, with significant CREM overexpression regardless of fusion type, when compared to either meningioma alone, or a larger group of CNS tumors.

CONCLUSIONS: This work provides further characterization of ICMT as an important CNS mesenchymal neoplasm that is prone to tumor recurrence, showing 2 prognostically relevant methylation subclasses, and warranting diagnostic distinction from other epigenetically and histologically related tumors. ICMTs show substantial overexpression of the CREM gene, independent of fusion type.

Basal forebrain cholinergic neurons (BFCNs) densely innervate auditory cortex (ACtx), conveying signals linked to internal brain states and external sensory cues. Acetylcholine (ACh) is known to rapidly modulate cortical circuits through nicotinic ACh receptor (nAChR)-mediated activation of layer 1 inhibitory neurons (L1-INs). However, BFCN terminals are also abundant in deeper layers, where their functional impact has received less attention. Using multi-plex in situ labeling across cortical layers and cell types, we found that layer 6 pyramidal neurons (L6-PNs) are highly enriched in diverse transcripts for nAChR subunits and muscarinic ACh receptors (mAChRs). In vivo optogenetic activation of BFCN axons revealed persistent modulation of regular spiking units in L2-6 but a rapid phasic activation only in L6. In acute slices, optogenetic activation of BFCN axons elicited fast nAChR-mediated excitatory post-synaptic potentials in L6-PNs, comparable to responses in L1-INs, and slower mAChR-mediated inhibitory responses. These findings identify L1-INs and excitatory L6-PNs as two major hubs for BFCN modulation of cortical circuits. By recruiting distinct receptor mechanisms and circuit motifs in L1 and L6, BFCNs may engage parallel pathways of cholinergic control that couple fast, transient modulation with slower, sustained regulation to shape cortical perception and plasticity.

BACKGROUND: The prevalence and role of frailty in preclinical Alzheimer's disease (AD) is unclear.

METHODS: Cross-sectional analyses of pre-randomization data from the Anti-Amyloid Treatment in Asymptomatic AD (A4) study were analysed to derive two models of a frailty index (FI)-full [FI-Full] and cognitive variables removed [FI-CVR]. The prevalence of frailty (FI > 0.25) according to amyloid status (Aβ+/-), and the association of frailty and cognition (determined by the Preclinical Alzheimer Cognitive Composite (PACC) score) and whether frailty moderates the relationship between amyloid status and cognition was assessed, adjusting for age, sex and education.

RESULTS: Four thousand four hundred eighty-six participants were included (mean age 71.3 ± 4.7 years, 30% participants Aβ+, 59% female). The prevalence of frailty in preclinical AD was 22% (or 44% when cognitive variables were removed from the FI). Using either FI model, in adjusted analyses, Aβ+ participants were more likely to be frail compared to Aβ- [FI-Full-Odds ratio (OR) 1.43 95% confidence interval (CI) 1.20-1.71, P < .001; FI-CVR-OR 1.21 95% CI 1.05-1.40, P < .008]. Frail participants had lower PACC scores compared to non-frail participants, on average (FI-Full-PACC score -0.58 95% CI -0.76 to -0.40, P < .001; FI-CVR-PACC score -0.26 95% CI -0.40 to -0.12, P < .001). Frailty did not influence the relationship between Aβ status and cognition.

CONCLUSIONS: In a cohort screened for a preclinical AD trial, elevated Aβ levels were associated with frailty and frailty was associated with reduced cognitive performance independent of elevated Aβ levels. These associations, and whether or not frailty is associated with longitudinal cognitive decline independent of Aβ status, warrant further study.

Therapeutic nucleic acid delivery has many potential applications, but it remains challenging to target extrahepatic tissues in a flexible and image-guided manner. To address this issue, we report a bioorthogonal pre-targeting strategy that uses focused ultrasound (FUS) to promote the delivery of mRNA-loaded lipid nanoparticles (mRNA-LNP). We synthesized amphiphilic click reactive anchors (ACRAs) consisting of a phospholipid PEG-conjugate functionalized with trans-cyclooctene (TCO) or its companion reactive partner methyltetrazine (mTz), producing ACRA-TCO and ACRA-mTz. ACRA derivatives were screened for cellular activity, yielding functionalized DOPE-PEG (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(polyethylene glycol)) derivatives outperforming those containing saturated lipid or branched PEG. Nanobubbles encapsulating ultrasound-responsive gas delivered ACRA-TCO to targeted cells and tissues using FUS. This pre-targeting promoted the subsequent delivery of mRNA-LNP functionalized with companion ACRA-mTz. Ultrasound pre-targeting enhanced the accumulation of mTz-functionalized nanoparticles in cell cultures and mice by 75% and up to 3.6-fold, respectively, and increased gene expression using mRNA-LNP in vivo. Microbubbles loaded with ACRA roughly doubled mRNA delivery to the heart, while ultrasound alone did not. Taken together, this report presents a modular, ultrasound-enabled strategy for enhancing nucleic acid delivery in targeted tissues.

PURPOSE OF REVIEW: Recent literature describes the deployment of different artificial intelligence (AI) technologies to potentially support infection prevention and control (IP&C) in both the community and healthcare environment. However, most studies focus on adults. This review explores the data and potential for AI to enhance IP&C for pediatric populations as well as recognizing important limitations.

RECENT FINDINGS: In community settings, AI can educate families about infections and risk, recognize potential clusters and outbreaks of infectious pathogens, and prescreen individually infected patients prior to entering a healthcare facility. For admitted patients, AI has been used to identify patients at risk for healthcare-associated infections (HAIs) such as central line associated blood stream infections, and may assist infection preventionists in abstracting chart data for HAI surveillance. Limitations include potential biases in training data and the lack of prospective studies validating the use of AI for IP&C purposes, especially in heterogeneous pediatric populations.

SUMMARY: AI can be a valuable tool in recognizing and controlling infections in both the community and healthcare settings. However, more studies in pediatric populations are needed, including prospective studies that validate tools created and trained on retrospective cohorts.

PURPOSE: The purpose of this study was to evaluate clinical factors that could differentiate patients with Sjogren's disease-related dry eye (SDDE) from non-SDDE.

METHODS: Patients with SDDE (n = 20), non-SDDE (n = 47), and healthy control participants (n = 29) were enrolled. Patients with dry eye were required to have physician-diagnosed dry eye disease for at least 6 months prior to the study. Sjogren's diagnosis was made according to 2016 Sjogren's disease criteria. Control participants had no known dry eye diagnosis or ocular and/or autoimmune disease. Several validated questionnaires regarding overall health and perception of health, mental status, review of systems, and ocular symptom were administered concurrently. Ocular surface and tear film parameters were evaluated using strict methodology.

RESULTS: Each questionnaire, each section within a questionnaire, and each question within a section were evaluated. Of systemic symptom questionnaires, only Profile of Fatigue and Discomfort Questionnaire (PROFAD) differentiated between SDDE and non-SDDE (10.5 vs. 5.8, P = 0.003). Among four Sjogren's Foundation recommended review of systems questions, only dry mouth differentiated patients with SDDE from non-SDDE (90% vs. 53%, P = 0.004). No mental health or ocular symptom questionnaires were different between SDDE versus non-SDDE. Among clinician-measured ocular surface parameters, only corneal fluorescein (2.5 vs. 1.0, P = 0.006) and conjunctival lissamine green (2.0 vs. 1.0, P = 0.001) staining scores were significantly worse in patients with SDDE compared to non-SDDE.

CONCLUSIONS: Ocular staining and dry mouth complaints were strongly associated with SDDE.

TRANSLATIONAL RELEVANCE: Ocular surface vital dye staining and inquiring about dry mouth complaints can be helpful to determine whether Sjogren's laboratory work-up should be considered in patients presenting with dry eye disease.

BACKGROUND: Patient-reported outcome measures (PROMs) are increasingly utilized in medial unicompartmental knee arthroplasty (UKA). Total knee arthroplasty studies have shown preoperative PROMs across multiple domains (knee physical function, general physical health, mental health, etc.) influence achievement of minimal clinically important difference (MCID) or patient acceptable symptom state (PASS). However, the utility of preoperative PROMs in medial UKA is underexplored. We investigated the impact of preoperative PROMs across multiple domains on MCID and PASS achievement in medial UKA.

METHODS: We retrospectively sourced 911 medial UKAs from a prospectively maintained multi-institutional arthroplasty registry. The following preoperative PROMs were recorded: 1) Knee Osteoarthritis Outcome Score-Physical Function Short Form (KOOS-PS); 2) Patient-Reported Outcomes Measurement Information System (PROMIS) Physical Function-10a (PF-10a); 3) PROMIS Global Physical; and 4) PROMIS Global Mental. These PROMs measure the following domains, respectively: 1) knee-specific physical function; 2) general physical function; 3) general physical health; and 4) general mental health. Multivariable forward binary logistic regression models identified factors independently associated with the achievement of literature-derived, anchor-based KOOS-PS thresholds for MCID and PASS.

RESULTS: Achievement of MCID was associated with a lower preoperative KOOS-PS score (P = 0.001) and a higher preoperative PROMIS Global Physical score (P = 0.013). Achievement of PASS was associated with a higher preoperative KOOS-PS score (P < 0.001), higher preoperative PROMIS Global Physical score (P = 0.038), higher preoperative PROMIS Global Mental score (P = 0.018), and higher preoperative PROMIS PF-10a score (P = 0.029).

CONCLUSION: Medial UKA patients who have severe knee symptoms have a high rate of MCID, but a low rate of PASS. Conversely, patients who have mild knee symptoms have a low rate of MCID, but a high rate of PASS. Higher baseline physical functioning, general physical health, and general mental health increase the likelihood of PASS achievement. Evaluating medial UKA patients who have several preoperative PROMs across multiple domains (knee physical function, general physical health, mental health, etc.) may aid counseling. Although PROMs are subjective and are not perfect measures, they are nevertheless validated and widely utilized, offering insights that can guide clinical decision-making. Preoperative PROMs should not restrict access to care.