Publications by Year: 2026

Assessing the severity and clinical impact of syndesmotic injury, especially when subtle, remains among the most diagnostically challenging conditions in foot and ankle orthopaedics. The distal tibiofibular syndesmosis is a complex, three-dimensional (3D) joint critical to ankle stability, and substantial injury to its ligament complex-especially if undetected or undermanaged-can result in long-term functional impairment. Unfortunately, routine clinical and conventional imaging evaluations are often inconclusive, falling short of capturing the dynamic and subtle nature of syndesmotic instability. Moreover, while myriad techniques have been reported to assist with diagnosis, none have proven reliably effective in isolation nor superior to comparison with the patient's uninjured side. Currently, there exists no singular test that can reproducibly function as a binary endpoint for diagnosis, suggesting that the future of accurate diagnosis-particularly of subtle instability, which can occur in various planes-will require either a composite testing endpoint, more sensitive 3D imaging modalities supported by artificial intelligence databases, or both. As orthopaedic practice advances towards individualised, precision-based care, diagnostic modalities must evolve to better reflect functional anatomy and quantify pathology in multiple planes at once. Weight-bearing computed tomography (WBCT) has emerged as a transformative tool that enables high-resolution, comparative 3D assessment of the syndesmosis under physiological load. Cadaveric and clinical studies have demonstrated WBCT's high diagnostic accuracy, reproducibility and efficiency-especially when integrated with automated 3D volumetric analysis. Quantitative thresholds such as side-to-side 3D volume differences or lateral translation measurements provide robust criteria for instability. Since it has been demonstrated that these imaging techniques are more accurate when comparing the injured to the uninjured limb rather than using population norms, the contralateral side can serve as an internal reference-improving precision and reducing reliance on generalised values. New machine learning algorithms using WBCT now allow real-time, high-accuracy evaluation with almost immediate comparative results. As normative databases grow, these algorithms will improve, clarifying variations between a patient's two ankles. Standardised software can also reduce measurement errors still common with conventional assessments. Despite its advantages, WBCT adoption is limited by cost and integration challenges. Nonetheless, its expanding applications represent a shift in orthopaedic diagnostics towards more functional, individualised imaging.

The NAIP-NLRC4 inflammasome senses pathogenic bacteria by recognizing the cytosolic presence of bacterial proteins such as flagellin and type III secretion system (T3SS) subunits. In mice, the NAIP-NLRC4 inflammasome provides robust protection against bacterial pathogens that infect intestinal epithelial cells, including the gastrointestinal pathogen Shigella flexneri. By contrast, humans are highly susceptible to Shigella, despite the ability of human NAIP-NLRC4 to robustly detect Shigella T3SS proteins. Why the NAIP-NLRC4 inflammasome protects mice but not humans against Shigella infection remains unclear. We previously found that human THP-1 cells infected with Shigella lose responsiveness to NAIP-NLRC4 stimuli, while retaining sensitivity to other inflammasome agonists. Using mT3Sf, a "minimal Shigella" system, to express individual secreted Shigella effector proteins, we found that the OspF effector specifically suppresses NAIP-NLRC4-dependent cell death during infection. OspF was previously characterized as a phosphothreonine lyase that inactivates p38 and ERK MAP kinases. We found that p38 was critical for rapid priming of NAIP-NLRC4 activity, particularly in cells with low NAIP-NLRC4 expression. Overall, our results provide a mechanism by which Shigella evades inflammasome activation in humans, and describe a mechanism for rapid priming of the NAIP-NLRC4 inflammasome.

IMPORTANCE: Pancreas divisum is implicated as an obstructive cause for acute pancreatitis. Observational data suggest endoscopic retrograde cholangiopancreatography (ERCP) with minor papillotomy reduces the risk of pancreatitis episodes. Even though this endoscopic procedure is widely used in practice, clinical trials are lacking.

OBJECTIVE: To determine whether ERCP with minor papillotomy reduces the risk of acute pancreatitis among adults with unexplained acute recurrent pancreatitis and pancreas divisum.

DESIGN, SETTING, AND PARTICIPANTS: This multicenter, sham-controlled, double-blind randomized clinical trial enrolled adults with 2 or more episodes of acute pancreatitis and pancreas divisum. Adults with other etiologies for acute pancreatitis or concomitant chronic calcific pancreatitis were excluded. The trial was conducted between September 1, 2018, and August 30, 2024, at 21 referral centers in the US and Canada. Last follow-up occurred on February 15, 2025.

INTERVENTION: Participants were randomized in a 1:1 ratio to ERCP with minor papillotomy or sham ERCP.

MAIN OUTCOMES AND MEASURES: The primary outcome was development of acute pancreatitis more than 30 days after randomization as a time-to-event outcome. The secondary outcomes included acute pancreatitis episode frequency and development of chronic calcific pancreatitis, diabetes, and exocrine pancreatic dysfunction.

RESULTS: A total of 148 participants were randomized (mean age, 54 [SD, 19.5] years; 68.2% female; 95.3% non-Hispanic or Latino and 87.2% White; mean lifetime acute pancreatitis episodes, 3 [SD, 2]; mean duct diameter, 2.2 [SD, 1.3] mm) and followed up for a median of 34 months (IQR, 21.7-45.7 months). Of the 75 participants in the ERCP with minor papillotomy group, 26 (34.7%) developed acute pancreatitis compared with 32 of 73 participants (43.8%) in the sham ERCP group (adjusted hazard ratio, 0.83 [95% CI, 0.49 to 1.41]). The incidence rate ratio for acute recurrent pancreatitis episode frequency was 0.25 (95% CI, 0.18 to 0.34) in the ERCP with minor papillotomy group vs 0.30 (95% CI, 0.23 to 0.41) in the sham ERCP group. There were no between-group differences in frequency and incidence of chronic calcific pancreatitis (4.0% in the ERCP with minor papillotomy group vs 2.7% in the sham ERCP group; risk difference [RD], 0.01 [95% CI, -0.05 to 0.07]), diabetes (15.8% vs 12.8%, respectively; RD, 0.03 [95% CI, -0.13 to 0.19]), and exocrine pancreatic dysfunction (7.7% vs 17.2%; RD, -0.10 [95% CI, -0.27 to 0.08]). The adverse event of acute pancreatitis within 30 days of randomization occurred more frequently in the ERCP with minor papillotomy group (14.7%) vs the sham ERCP group (8.2%) (RD, 0.06 [95% CI, -0.04 to 0.17]).

CONCLUSIONS AND RELEVANCE: Among patients with unexplained acute recurrent pancreatitis and pancreas divisum, ERCP with minor papillotomy does not reduce the risk of another episode of acute pancreatitis or related sequelae.

TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT03609944.

IMPORTANCE: It is unclear whether and the extent to which subclinical myocardial injury or stress coexisting with prediabetes is associated with the risk of heart failure (HF).

OBJECTIVE: To evaluate the joint associations of prediabetes and subclinical myocardial injury or stress with incident HF risk.

DESIGN, SETTING, AND PARTICIPANTS: This post hoc prospective cohort study analyzed data from the Systolic Blood Pressure Intervention Trial (SPRINT). Two analytic samples were used: (1) adults with hypertension without diabetes or prior HF for the baseline biomarkers analysis and (2) participants with biomarker measurements at both baseline and 12 months for the longitudinal biomarkers' change. Prediabetes was defined as a fasting plasma glucose level of 100 to 125 mg/dL. Subclinical myocardial injury was defined as a high-sensitivity cardiac troponin I (hs-cTnI) level of 6 ng/L or higher in men and 4 ng/L or higher in women and subclinical myocardial stress defined as an N-terminal pro-B-type natriuretic peptide (NT-proBNP) level of 125 pg/mL or higher. A 25% or greater increase in any biomarker concentration from baseline to 12 months defined longitudinal change. Data were analyzed between January 1 and May 31, 2025.

MAIN OUTCOMES AND MEASURES: The primary outcome was adjudicated incident HF. Cox proportional hazards models were used to estimate hazard ratios (HRs) for HF across joint categories of prediabetes and biomarker elevation.

RESULTS: Of 8234 participants (mean [SD] age, 68 [9] years; 37.1% women), 3271 (39.7%) had prediabetes, 2942 (35.7%) had subclinical myocardial injury, and 3593 (43.6%) had subclinical myocardial stress. Over a median follow-up of 3.2 years (IQR, 2.8-3.8 years), 122 participants developed HF. Compared with normoglycemia and no myocardial injury, those with both prediabetes and injury had the highest HF risk (HR, 4.20; 95% CI, 2.31-7.63); similar findings were observed for myocardial stress (HR, 5.20; 95% CI, 2.52-10.70). In the longitudinal analysis (median follow-up, 2.3 years [IQR, 1.9-2-8 years]), 7449 participants with both prediabetes and a 25% or greater increase in hs-cTnI or NT-proBNP level had the highest risk of HF (for hs-cTnI: HR, 3.05; 95% CI, 1.58-5.88; for NT-proBNP: HR, 2.39; 95% CI, 1.28-4.46).

CONCLUSIONS AND RELEVANCE: These findings suggest that among adults with hypertension, prediabetes in combination with subclinical myocardial injury or stress is associated with a significantly elevated risk for HF. These findings support the integration of glycemic status and cardiac biomarkers profiling to improve HF risk stratification and guide prevention.

INTRODUCTION: Low engagement with tobacco treatment by Black and/or Hispanic adults with cancer who smoke may perpetuate health disparities. The current secondary analysis of a tobacco treatment trial aimed to describe the racial and ethnic characteristics of Black and/or Hispanic cancer patients and to examine the association between these characteristics and treatment outcomes.

METHOD: Eligibility criteria were: 18 + years old, newly diagnosed with cancer, smoked ≥ 1 cigarette in the past month, and spoke English or Spanish. Participants who self-identified as Black and/or Hispanic (N = 41), or White (N = 255) received (1) four telephone counseling sessions and information about smoking cessation medication or (2) eight telephone counseling sessions and 3 monthly booster sessions plus 12 weeks of free smoking cessation medication. We examined sociodemographic characteristics, smoking history, and cultural characteristics (e.g., racial/ethnic identity, language, familism, religiosity) at baseline and smoking abstinence at 3- and 6-month follow-ups.

RESULTS: Black and/or Hispanic patients smoked fewer cigarettes per day (10.5 vs. 14.7 cigarettes per day, p = .03) and reported higher levels of baseline quitting self-efficacy compared to White patients (6.5 vs. 5.3, p = .006). Black and/or Hispanic patients had similar treatment engagement, protocol adherence, and biochemically confirmed smoking abstinence at 3- and 6-month follow-ups compared to White non-Hispanic patients, controlling for the treatment group.

DISCUSSION: Findings suggest that virtual tobacco treatment was accessible and effective for Black and/or Hispanic cancer patients. Tailored treatment may support greater benefits from tobacco treatment among Black and/or Hispanic patients who experience a greater health burden associated with smoking.

BACKGROUND: Artificial intelligence (AI) and machine learning (ML) are increasingly integrated into healthcare, offering potential advancements in patient-reported outcome measures (PROMs) for surgical populations. Improved PROMs can enhance patient-centered care by accurately capturing patient experiences with minimal burden.

OBJECTIVE: In the context of surgery, where recovery trajectories vary widely, this study aims to systematically review the use of AI and ML in the development, application, and prediction capabilities of PROMs in surgical populations, with a focus on psychometric properties and the predictive accuracy of post-surgical outcomes.

METHODS: A comprehensive search of the PubMed database was conducted from inception until August 2024. Studies were included if they utilized AI or ML in the development, application, or predicting PROMs for surgical patients. Methodological quality was assessed using COSMIN and PROBAST tools, depending on study design. A qualitative synthesis of findings was performed.

RESULTS: Twenty-two studies met the inclusion criteria, with 19 rated as high quality. Six studies focused on developing computer adaptive tests (CAT) PROMs, seven on evaluating psychometric properties, and five on ML for post-surgical outcome prediction. CAT PROMs showed comparable measurement accuracy to traditional PROMs, good to excellent construct validity, and significantly reduced patient burden by reducing the length of questionnaires. ML algorithms, such as logistic regression, random forests, extreme gradient boosting, and neural networks, achieved similar predictive accuracy for post-surgical outcomes, with no single model demonstrating consistent superiority.

CONCLUSIONS: AI and ML have the potential to improve PROM utilization in surgical care by enhancing efficiency and personalization while maintaining data quality. Clinicians can use AI-driven PROMs to reduce patient burden and integrate ML models for accurate post-surgical outcome prediction, thereby optimizing patient-centered care.

INTRODUCTION: Rehabilitation at inpatient rehabilitation facilities (IRFs) has been shown to improve seriously injured patients' long-term functional independence. However, not all patients with severe injuries are discharged to an IRF. The aim of this study is to examine differences in the proportion of severely injured patients discharged to IRFs across US states and assess the association between the regional supply of IRFs and the likelihood of IRF discharge.

METHODS: We performed a retrospective analysis of 2021 Healthcare Cost and Utilization Project State Inpatient Databases across 13 states. We included severely injured (ISS >15) adult patients who survived to hospital discharge. We calculated the marginal probability of discharge to IRF, skilled nursing facility, home health agency, or home without services across states using a multinomial logistic regression model to control for patient demographics, insurance type, injury severity, medical comorbidities, and trauma center level. We also performed a mixed-effects logistic regression to evaluate the association between the supply of IRFs within individual states with the likelihood of discharge to IRF.

RESULTS: We identified 104,017 severely injured patients. Across all 13 included states, 13% of patients were discharged to IRFs. There was considerable variation in the adjusted probability of discharge to an IRF across states, ranging from 6.4% in Oregon (95% confidence interval [CI], 5.6-7.1%) to 22.1% in Kentucky (95% CI, 20.8-23.4%), after controlling for potential confounders. The state-level supply of IRFs ranged from 0.49 to 8.63 per 1,000,000 population in Maryland and Arkansas, respectively. Each additional IRF per 1,000,000 population was associated with 11% increased odds of discharge to IRF (95% CI, 1.01-1.21; p = 0.024).

CONCLUSION: Severely injured patients face substantial differences in accessing high-level rehabilitation care at an IRF depending on their state of residence. Increasing the availability of IRFs within underserved states may improve access to specialized rehabilitation care for trauma patients.

LEVEL OF EVIDENCE: Therapeutic/Care Management; Level III.

The role of inflammation in cortical superficial siderosis (cSS), a marker of cerebral amyloid angiopathy (CAA) linked to high hemorrhage risk, is unclear. We examined 15 patients with cSS using 3 T post-contrast vessel wall MRI (VWI) and CSF analysis. Although only 27% met current CAA-ri criteria, 93% showed vessel wall enhancement or sulcal hyperintensities near cSS, frequently extending beyond. Seven patients with follow-up VWI demonstrated corticosteroid-responsive regression of inflammation. CSF albumin quotients, indicating blood-brain barrier dysfunction, correlated with MRI inflammation scores. These findings reveal subclinical meningovascular inflammation in cSS and support VWI for detecting a broader CAA-related inflammation spectrum.

The bispecific antibody tarlatamab recruits T cells to cancers expressing the neuroendocrine epitope DLL3. Tarlatamab is effective in small cell lung cancer (SCLC), but clinical outcomes vary, and no biomarkers enable patient selection. Single-cell RNA sequencing of SCLC biopsies identifies heterogeneity in DLL3 expression, and analysis of circulating tumor cells (CTCs) distinguishes individual patients as predominantly DLL3Pos or DLL3Low. In a prospective cohort of 20 patients, pretreatment DLL3 expression on CTCs predicts tarlatamab clinical benefit (85% sensitivity, 100% specificity). Necrotic CTC clusters in blood accompany treatment-induced tumor lysis. Acquired resistance to tarlatamab is associated in some cases with loss of DLL3 expression, but persistence of other targetable neuro-endocrine epitopes; in other patients, DLL3 is retained on CTCs, but accompanied by systemic markers of T cell dysfunction. Quantitation of DLL3-positive CTCs identifies patients likely to benefit from tarlatamab, and longitudinal monitoring may guide therapeutic decision-making at the time of acquired resistance.