Publications

BACKGROUND: Diagnosing cardiac amyloidosis (CA) on echocardiography can be challenging due to the imaging overlap between CA and more prevalent causes of a hypertrophic phenotype. This study sought to (1) evaluate the performance of artificial-intelligence (AI) derived measurements incorporated into the established multiparametric echocardiographic scoring system to detect CA; (2) develop and validate an AI-based deep-learning model for video-based detection of CA on echocardiography.

METHODS: The study population comprised 5776 patients (CA, 2756; controls, 3020). The training data set included patients from the UK National Amyloidosis Center and Taiwan MacKay Memorial Hospital (CA, 2241; controls, 2130). External test data sets were obtained from the US Duke University Health System (CA, 334; LVH controls, 668) and Japan National Cerebral and Cardiovascular Center (CA, 181; LVH controls, 222).

RESULTS: The multiparametric echocardiographic score computed using AI-derived measurements achieved an accuracy of 79.5% (sensitivity, 75.4%; specificity, 81.5%) in the United States cohort and 79.7% (sensitivity, 81.6%; specificity, 78.1%) in the Japan cohort. The deep-learning model demonstrated accuracies of 96.2% (sensitivity, 96.8%; specificity, 95.7%) and 95.8% (sensitivity, 97.3%; specificity, 94.3%) in the internal validation and internal test sets, respectively. External validation of the deep-learning model showed accuracies of 87.5% (sensitivity, 86.6%; specificity, 87.9%) in the United States and 88.4% (sensitivity, 92.3%; specificity, 85.3%) in the Japanese cohort. Subgroup analysis demonstrated that the deep-learning model showed robust discrimination of CA from other hypertrophic phenocopies: CA versus hypertension (area under the curve [AUC], 0.92 [95% CI, 0.91-0.94]), CA versus hypertrophic cardiomyopathy (AUC, 0.91 [95% CI, 0.87-0.94]), CA versus aortic stenosis (AUC, 0.93 [95% CI, 0.90-0.95]), CA versus chronic kidney disease (AUC, 0.93 [95% CI, 0.91-0.95]). The deep-learning model was able to classify a greater proportion of patients compared with the AI-derived multiparametric echocardiographic score and achieved superior diagnostic accuracy (AUC, 0.93 [95% CI, 0.91-0.95] versus AUC, 0.88 [95% CI, 0.85-0.90]; P<0.001).

CONCLUSIONS: Both the multiparametric echocardiographic score computed from AI-derived measurements and the fully automated deep-learning model can accurately identify patients with CA in globally diverse cohorts, with the deep-learning model providing superior performance.

Somatic mutations in TET2 drive hyper-inflammation in clonal hematopoiesis of indeterminate potential (CHIP), but the molecular link between TET2 inactivation and myeloid immune activation remains unclear. We used in vivo genome-wide genetic perturbations enabled by ultra-diverse barcoding in primary wild-type (WT) or Tet2 knockout (KO) Cas9+ hematopoietic stem-progenitor cells (HSPCs) to elucidate the basis of Tet2 KO inflammation. We uncover a metabolic circuit by which Tet2 restrains O-linked N-acetylglucosamine (O-GlcNAc) glycosyltransferase (Ogt), a Tet2 binding partner and metabolic sensor. Tet2 loss disrupts this inhibitory Tet2-Ogt interaction, and dysregulated Ogt facilitates widespread H3K4 trimethylation including lipid-related gene loci and inflammatory lipid droplet formation. We identified that ATP citrate lyase (Acly) is decorated with O-GlcNAc and is a critical node for lipid accumulation and inflammation in Tet2 KO. These findings reveal that Tet2 suppresses inflammation by gating nutrient-responsive chromatin remodeling and nominate metabolic interventions to restrain inflammatory disease in TET2-mutant clonal hematopoiesis.

BACKGROUND: Gestational diabetes mellitus (GDM) increases offspring obesity risk, but whether this occurs via changes in human milk composition, including alterations in human milk oligosaccharides (HMOs), is unknown.

OBJECTIVES: This study aimed to identify differences in HMO concentrations in mothers with and without GDM and test whether GDM-associated HMOs are associated with infant growth, body composition, and fecal microbiome characteristics over the first 6-mo of life.

METHODS: Human milk was collected at 1-mo postpartum from 337 females (49 with GDM) who fed their infants breastmilk exclusively. HMOs were quantified by high-performance liquid chromatography and multivariate regression models were used to test differences in HMO concentrations by GDM status (false discovery rate adjustment for multiple testing set at q < 0.05). HMOs associated with GDM were then tested for associations with infant growth, body composition, and 1 and 6-mo infant fecal microbial abundances measured by metagenomic whole-genome sequencing.

RESULTS: Participants with GDM had ∼1 SD higher milk 6'sialyllactose (6'SL) {[β (95% confidence interval): 0.58 (0.20, 0.96)] and lacto-N-fucopentaose III (LNFP III) III [95% CI: 0.55 (0.16, 0.94)]} compared with those without GDM and 6'SL concentration was also positively associated with weight and length gain. Although infants of mothers with GDM had lower 1-mo fecal α-diversity and altered abundances of 6 of 56 microbial species detected compared with those without GDM, microbial features were not associated with the concentration of either 6'SL or LNFP III and evidence for mediation of GDM-growth and GDM-microbiome by HMOs was not found.

CONCLUSIONS: Mothers with a GDM diagnosis had higher milk concentrations of LNFP III and 6'SL, and 6'SL was in turn associated with increased infant growth rate, but neither HMO was associated with differential infant gut microbial abundances. The results suggest that the link between 6'SL and faster infant growth, if causal, occurs via mechanisms independent of the infant gut microbiome. This study was registered at clinicaltrials.gov as NCT03301753.

Our studies have established that loss-of-function mutations in the Wnt signaling antagonist SFRP4 (Secreted Frizzled Related Protein 4) cause Pyle disease, a rare skeletal disease characterized by limb deformity. Pyle disease can also present with a number of dental conditions, including tooth decay and taurodontism. Aside from the sparse clinical descriptions of the tooth anomalies seen in Pyle disease, the role of Sfrp4 in teeth has not been investigated. Here we show that in adult mouse incisors, Sfrp4 is expressed in the mesenchymal and epithelial compartments and their derivatives as well in the developing apex of the molar roots, pulp, odontoblasts, and periodontal ligament. We report that Sfrp4 deletion in mice leads to markedly shorter incisors, with reduced dentin mineral apposition and enamel volume. In addition, we find that Sfrp4 deletion leads to a reduction in the root length and bifurcation height of the developing molars, both features of taurodontism. Rodent incisors grow continuously during the animal's lifetime, thanks to the maintenance and interactions of epithelial (EpSCs) and mesenchymal stem cells (MSCs) at their apex. As such, the mouse incisor is a powerful tool for studying adult stem cells, their interactions, and their regulation. Using this model, we found that Sfrp4 deletion leads to a significant decrease in Gli1+ MSCs and in Ki67 levels in mesenchymal transient amplifying cells and preameloblasts. Sfrp4-null incisors have reduced type I collagen levels and altered amelogenin and Mmp20 secretion accompanied by changes in the enamel maturation stages. When incisor growth is accelerated by preventing occlusion through tooth clipping, Sfrp4-null incisors grow significantly slower than wt incisors do. Our study suggests a key role for Sfrp4 in the fast-growing teeth (incisors) and developing molars, where there is a need to balance the maintenance and differentiation of the stem cell niche.

Bone density and microarchitecture measurements from high-resolution peripheral computed tomography (HR-pQCT) are increasingly being used to predict fracture risk and to gain insight into the pathophysiology of skeletal fragility. Using the first-generation HR-pQCT scanner, we previously showed that extraosseous soft tissue can impact HR-pQCT measurements. Yet similar data is not available for the second-generation scanner. Thus, we aimed to determine the impact of increased soft tissue on bone density, microarchitecture and strength measurements acquired using the second-generation HR-pQCT scanner. We performed HR-pQCT scans on a hydroxyapatite phantom and in human volunteers (n = 12) with no soft tissue covering, and with a thin (0.5 cm) and thick (1 cm) layer of soft tissue surrounding the phantom or limb. We found that density values of the phantom were minimally affected by the thin layer of soft tissue. In contrast, with the thick (1 cm) layer of soft-tissue, bone density was significantly lower than baseline (no bolus), with greater deficits as the density of the rod increased (-1% to -3.6%, p < 0.01 for all). In human volunteers, soft tissue layering influenced measures of both cortical and trabecular microarchitecture at the distal tibia and radius, with larger differences observed in trabecular versus cortical measures and at the tibia compared to the radius. For example, at the tibia, Tt.BMD was lower than the baseline scan under both soft tissue layering conditions (thin: -1.3%, p < 0.001; thick: -2.4%, p = 0.003), while at the radius Tt.BMD was only significantly lower for the thick layer (-1.2%, p = 0.004). Ct.BMD followed a similar pattern, with slightly greater magnitude of BMD decline with increased soft tissue (thin: -1.5%, p < 0.001, thick -2.8%, p < 0.001 at the tibia) compared to Tt.BMD. Altogether our results indicate that HR-pQCT measurements at both the metaphyseal and diaphyseal sites must be interpreted carefully when comparing subjects with varying body composition, or when assessing longitudinal changes in individuals who experience marked changes in weight and/or body composition as true differences to these measures may be more or less extreme than they appear.

Nature employs post-translational modifications (PTMs) to induce proximity between proteins by engendering new interactions. Furthermore, we find that protein ligands are invariably proximal to a lysine. Inspired by these two observations, we developed group-transfer chimeras (GRCs) that append a moiety-of-interest to the lysine side chain. GRCs employ a protein's ligand and a handle with a transferase-type reactivity to modify the proximal lysine. Contemporary lysine-targeting group-transfer handles were incompatible with GRCs due to their hydrolytic instability, large size, high reactivity, and synthetic incompatibility with diverse ligands. Accordingly, we developed an N-(sulfonyl)-N-(trifluoroethyl)-ethanamide (SuFA) handle that is stable, small, and exhibits tunable reactivity and synthetic compatibility with diverse ligands and proteins. Using GRCs that group-transfer binders of tags (e.g., HaloTag, FKBP) onto proteins overexpressed in cancer cells, we displayed these binders on the surface of the cancer cell. With a universal T cell engager (UniTE) that binds to the displayed ligands and T cells, these GRCs induced proximity between cancer cells and cytotoxic T cells, leading to the latter's activation. We envision the GRC platform to find utility in basic research and biomedicine.

BACKGROUND: Since 2013, fentanyl has dominated unregulated drug supplies in the northeast and has become a key driver of overdose deaths. In this analysis, we sought to identify the characteristics associated with fentanyl use frequency among people who use drugs in Rhode Island.

METHODS: We examined baseline data from the Rhode Island Prescription and Illicit Drug Study, which enrolled participants from August 2020 to February 2023. We used an ordinal logistic regression to identify sociodemographic and drug use characteristics that were associated with prior month fentanyl use frequency, which was categorized as none, less than weekly, or at least weekly.

RESULTS: Among 471 participants, 210 (44%) reported no prior month use of fentanyl or drugs containing fentanyl, 121 (26%) reported less than weekly use, and 140 (30%) reported at least weekly use. Of those who reported at least weekly use, 74% reported intentional use. In bivariate analyses, the proportion of participants utilizing harm reduction practices (e.g., fentanyl test strip use, naloxone carriage) increased with fentanyl use frequency in a dose-response manner. In adjusted analyses, prior month injection drug use, fentanyl preference, regular use of crack cocaine, lifetime overdose history, and current enrollment in opioid agonist therapy were significantly associated (p< 0.05) with greater fentanyl use frequency.

CONCLUSIONS: Fentanyl exposure and intentional fentanyl use were common in the sample. Our findings suggest a need for increased investment in community-based harm reduction services and low-barrier, patient-centered treatment for polysubstance use, including those involving stimulant use disorders.

BACKGROUND: Enteral nutrition (EN) delivery is often interrupted in the intensive care unit (ICU), and while continuous 24-hour feeding is standard practice, emerging evidence from circadian biology and pilot trials suggests that daytime-restricted EN may enhance nutritional adequacy and patient outcomes by aligning feeding with biological rhythms.

METHODS: This quality improvement study describes a novel, standardized daytime-restricted EN protocol in a community hospital ICU and retrospectively evaluate its real-world implementation. The protocol involved a stepwise transition in EN delivery, beginning with continuous trophic feeding (acute/initial phase), followed by daytime-restricted 12-hour cyclic feeding (anabolic recovery phase), and advancing to intermittent daytime-restricted feeding (chronic recovery phase). A convenience sample of 22 adult ICU patients (12 received continuous 24-hour EN; 10 with the daytime-restricted EN protocol) was analyzed. Clinical data were extracted from electronic medical records, including EN infusion rates, duration, and interruptions.

RESULTS: Patients in the daytime-restricted group received EN at higher infusion rates (median 87.5 vs. 40.0 mL/hr), over fewer hours per day (11.0 vs. 14.5 hours), experienced fewer interruptions (1.0 vs. 9.5 hours/day), and received a greater percentage of their prescribed nutritional volume (90.0% vs. 57.5%) compared to the continuous group (all P value < 0.05). Vomiting was more frequently reported in the daytime-restricted group, while constipation was more common in the continuous group, though these differences were not statistically significant.

CONCLUSION: This preliminary evaluation supports the feasibility of implementing a daytime-restricted EN protocol in an adult ICU and suggests potential advantages in delivery consistency and nutritional adequacy. To support broader implementation, larger prospective studies across broader ICU populations are necessary.

C-X3-C motif chemokine ligand 1 (CX3CL1), a structurally unique chemokine in the central nervous system (CNS), shapes physiological and pathological processes via specific binding to its receptor, C-X3-C motif chemokine receptor 1 (CX3CR1). Empirical evidence indicates that this signaling axis exerts dual neuroinflammatory effects: It restrains microglial hyperactivation, yet can promote inflammation under conditions such as chronic stress. Notably, it preserves synaptic plasticity and facilitates remyelination. Age-associated reductions in CX3CL1 exhibit a strong correlation with cognitive decline; administration of exogenous CX3CL1 partially mitigates these deficits. This study provides a comprehensive account of the multifaceted functions and regulatory mechanisms of CX3CL1 in CNS diseases, thereby establishing a basis for potential new therapeutic targets.

The substantial adverse effects associated with opioid-based analgesia and its contribution to postoperative dependence have prompted a shift toward multimodal, opioid-sparing perioperative strategies. Non-opioid analgesics now form the cornerstone of contemporary perioperative management and Enhanced Recovery After Surgery (ERAS) pathways. This review synthesizes current evidence on the efficacy, safety, and clinical utility of major non-opioid analgesic classes, including nonsteroidal anti-inflammatory drugs (NSAIDs), acetaminophen, N-methyl-D-aspartate (NMDA) antagonists, intravenous lidocaine, gabapentinoids, α2-agonists, regional anesthesia techniques, and glucocorticoids, for postoperative pain management. Evidence from randomized controlled trials, systematic reviews, and meta-analyses was evaluated with emphasis on analgesic effectiveness, opioid-sparing capacity, recovery outcomes, and adverse effects. The efficacy of non-opioid multimodal analgesia (MMA) stems from the synergistic targeting of distinct pain pathways. Across drug classes, non-opioid agents demonstrate clinically meaningful opioid-sparing effects while providing analgesia that is comparable to opioid-based regimens. Acetaminophen and NSAIDs are cornerstones of MMA, supported by extensive high-quality evidence. NMDA antagonists such as ketamine show particular benefit in major surgical procedures for modulating central sensitization and preventing chronic pain, whereas intravenous lidocaine has unique advantages in accelerating gastrointestinal recovery and may reduce hospital length of stay. Gabapentinoids may serve as adjuncts but exhibit heterogeneous efficacy and a side-effect profile (e.g., sedation) that necessitates selective use. Preoperative α2-agonists consistently prolong analgesia and reduce perioperative opioid requirements. As a core component of MMA, regional anesthesia techniques demonstrate robust reductions in both prolonged postoperative opioid use and the incidence of chronic postsurgical pain, and perioperative glucocorticoids such as dexamethasone contribute potent dual analgesic and antiemetic effects. The collective evidence indicates that MMA, integrating agents with complementary mechanisms, provides superior pain control, enhanced functional recovery, and meaningful reductions in opioid exposure. Broader implementation of standardized, procedure-specific multimodal protocols may further decrease opioid-related harms and strengthen alignment with ERAS principles. Future research should prioritize long-term outcomes and optimization of multimodal combinations to advance the transition toward a post-opioid paradigm in surgical care.