About

Yuan Wu is an Associate Professor in Biostatistics & Bioinformatics at Duke University and a member of the Duke Cancer Institute. He is based at the Duke Department of Biostatistics and Bioinformatics, located at 2424 Erwin Road, Durham, NC. His professional role involves research and teaching within the division of Biostatistics, contributing to the academic and clinical research community at Duke.

Research topics

• Physics
• Optics
• Nuclear physics
• Materials science
• Atomic physics

Selected publications

• Causal relationship between plasma metabolites and hypertension: A Mendelian randomization study

  Medicine · 2025-10-03

  articleOpen access

  Observational studies have indicated an association between metabolites and hypertension. However, establishing a causal relationship remains challenging. This study aims to investigate this causal relationship using Mendelian randomization methodology. Genome-wide association study summary statistics for plasma metabolites were obtained from 3 studies, encompassing over 2000 varieties. Genome-wide association study summary statistics for hypertension were derived from 2 studies, with a sample size of 122,996 and 129,909, respectively. The main method used was the inverse variance weighting (IVW) method, and the results were adjusted using Bonferroni multiple correction. The IVW method estimated effects between the levels of taurochenodeoxycholate, 1-dihomo-linolenoyl-GPC (20:3n3 or 6), N-acetylphenylalanine, and dihomo-linolenate (20:3n3 or n6) and hypertension are 1.041 (95% confidence interval [CI] 1.022-1.059, P = 1.49e-05), 1.023 (95% CI 1.015-1.031, P = 5.32e-08), 1.029 (95% CI 1.021-1.037, P = 6.10e-12), and 1.019 (95% CI 1.010-1.028, P = 1.54e-05), respectively. The IVW method estimated effects between the levels of 2-hydroxyoctanoate, 1-docosapentaenoyl-GPC (22:5n3), X-11538, gamma-glutamyl-alpha-lysine, and 2-butenoylglycine and hypertension are 0.954 (95% CI 0.936-0.972, P = 8.87e-07), 0.981 (95% CI 0.973-0.989, P = 4.14e-06), 0.981 (95% CI 0.975-0.987, P = 7.53e-10), 0.942 (95% CI 0.922-0.963, P = 8.71e-08), and 0.930 (95% CI 0.902-0.958, P = 2.22e-06), respectively. In summary, there are indications that some metabolites have a causal relationship with hypertension. It is necessary to conduct further investigations to comprehend the underlying biological mechanisms that support these associations.

  PublisherDOI

• miR-107-enriched exosomes promote ROS/wnt/autophagy, inhibit intracellular mycobacterial growth and attenuate lung infection

  Frontiers in Immunology · 2025-07-04 · 2 citations

  articleOpen access

  ) infection remain poorly characterized. Here, we found that miR-107 highly expressed in exosomes from plasma of TB patients but not healthy control (HC) subjects. Consistently, such miR-107-high exosomes were also detected in both the extracellular fluid released by mycobacterial-infected macrophages and the plasma of mycobacterial-infected mice. Interestingly, adding the miR-107-high plasma exosomes or the miR-107 mimics to infected THP-1 macrophages inhibited intracellular mycobacterial growth. Consistently, while nanoscale and fluorescence imaging revealed that miR-107 could be transferred inter-cellularly via exosomes, miR-107-enriched exosomes from miR-107 overexpressing cells also inhibited mycobacterial growth in THP-1 macrophages and primary monocytes/peripheral blood mononuclear cells (PBMC). Mechanistically, miR-107-high exosomes increased ROS production; miR-107 regulated Wnt pathway by targeting Wnt16 and promoted autophagy in THP-1 macrophages. Furthermore, treatment of infected mice with miR-107-enriched exosomes reduced mycobacterial infection in lung tissues. Our results raise a possibility to explore miR-107-high plasma exosomes for a potential surrogate marker for TB. Findings suggest that exosomes enriched with miR-107 or other bio-active molecules may potentially serve as an attractive approach for treatment of infection.

  PublisherOA PDFDOI

• Genetic ancestry concordant RNA splicing in prostate cancer involves oncogenic genes and associates with recurrence

  npj Precision Oncology · 2025-01-29 · 2 citations

  articleOpen access

  Black men suffer disproportionately from prostate cancer (PCa) compared to men of other races and ethnicities. Comparing the molecular landscape of PCa among Black and White patients has the potential to identify targets for development of new precision medicine interventions. Herein, we conducted transcriptomic analysis of prostate tumors and paired tumor-adjacent normals from self-reported Black and White PCa patients and estimated patient genetic ancestry. Clinical follow-up revealed increased biochemical recurrence (BCR) among Black patients compared to White patients with high-grade PCa. Transcriptomic analysis identified differential alternative RNA splicing events (ARSs) between Black and White PCa patients. Genes undergoing genetic ancestry-concordant ARSs in high-grade or low-grade tumors involved cancer promoting genes. Most genes undergoing genetic ancestry-concordant ARSs did not exhibit differential aggregate gene expression or alternative polyadenylation. A number of the genetic ancestry-concordant ARSs associated with BCR; thus, genetic ancestry-concordant RNA splice variants may represent unique targets for PCa precision oncology.

  PublisherOA PDFDOI

• Self-Management Task Performance and Its Association with Biomarkers in Elderly Patients with Chronic Heart Failure: A Cross-Sectional Analysis

  Patient Preference and Adherence · 2025-09-01 · 1 citations

  articleOpen accessSenior authorCorresponding

  Purpose: This study aimed to investigate the relationship between self-management task performance and key heart failure biomarkers (NT-proBNP and uric acid) in elderly patients with chronic heart failure (CHF), and to explore the mediating role of disease severity in this relationship. Design Patients and Methods: A total of 103 elderly CHF patients were recruited. And cross-sectional study conducted at two tertiary hospitals. Self-management task performance was assessed using a validated six-task evaluation scale. Clinical data, including NT-proBNP, uric acid, and other laboratory markers, were collected. Participants were stratified into low and high task performance groups based on the median score. Data were analyzed using partial correlation analysis, logistic regression, and restricted cubic spline analysis. Results: Patients with higher task performance were significantly younger (p < 0.001). Task performance was negatively correlated with NT-proBNP (r = -0.337, p < 0.001) and uric acid (r = -0.279, p = 0.005), indicating that lower performance was associated with higher biomarker levels, reflecting more severe disease progression. A non-linear decreasing trend in both NT-proBNP and uric acid levels was observed as task performance increased. Among the self-management tasks, medication-related tasks had the lowest success rates, while tasks related to oedema and dietary management showed higher performance rates. Conclusion: Lower self-management task performance is significantly associated with elevated NT-proBNP and uric acid levels in elderly CHF patients, suggesting that poorer performance may indicate more advanced disease progression. These findings highlight the importance of targeted interventions to enhance self-management skills, particularly in medication adherence and disease monitoring, to improve patient outcomes. Future research should investigate the long-term clinical impact of improving task performance in CHF patients.

  PublisherDOI

• Improving Variance and Confidence Interval Estimation in Small-Sample Propensity Score Analyses: Bootstrap vs. Asymptotic Methods

  ArXiv.org · 2025-11-14

  preprintOpen access

  Propensity score (PS) methods are widely used to estimate treatment effects in non-randomized studies. Variance is typically estimated using sandwich or bootstrap methods, which can either treat the PS as estimated or fixed. The latter is thought to be conservative. Comparisons between the sandwich and bootstrap estimators have been compared in moderate to large sample sizes, favoring the bootstrap estimator. With the growing interest in treatments for rare disease and externally controlled clinical trials, very small sample sizes are not uncommon and the asymptotic properties of sandwich estimators may not hold. Bootstrap methods that allow for PS re-estimation can also generate problems with quasi-separation in small samples. It is unclear whether it is safe to prefer sandwich estimators or to assume that treating the PS as fixed is conservative. We conducted a Monte Carlo simulation to compare the performance of bootstrap versus sandwich variance and CI estimators for average treatment effects estimated with PS methods. We systematically evaluated the impact of treating the PS as fixed versus re-estimating it. These methodological comparisons were performed using Inverse Probability of Treatment Weighting (IPTW) and Augmented Inverse Probability of Treatment Weighting (AIPW) estimators. Simulations assessed performance under various conditions, including small sample sizes and different outcome and treatment prevalences. We illustrate the differences in our motivating example, the LIMIT-JIA trial. We show that the sandwich estimators can perform quite poorly in small samples, and fixed PS methods are not necessarily conservative. A stratified bootstrap avoids quasi-separation and performs well. Differences were large enough to alter statistical conclusions in our motivating example, LIMIT-JIA.

  PublisherOA PDFDOI

• Group Sequential Test for Two‐Sample Ordinal Outcome Measures

  Statistics in Medicine · 2025-03-15 · 1 citations

  articleOpen access1st authorCorresponding

  Group sequential trials include interim monitoring points to potentially reach futility or efficacy decisions early. This approach to trial design can safeguard patients, provide efficacious treatments for patients early, and save money and time. Group sequential methods are well developed for bell-shaped continuous, binary, and time-to-event outcomes. In this paper, we propose a group sequential design using the Mann-Whitney-Wilcoxon test for general two-sample ordinal data. We establish that the proposed test statistic has asymptotic normality and that sequential statistics satisfy the assumptions of Brownian motion. We also include results of finite sample simulation studies that show our proposed approach has the advantage over existing methods for controlling Type I errors while maintaining power for small sample sizes. A real data set is used to illustrate the proposed method and a sample size calculation approach is proposed for designing new studies.

  PublisherOA PDFDOI

• Spatial Configuration Mechanism of Rural Tourism Resources Under the Perspective of Multi-Constraint Synergy: A Case Study of the Nujiang Dry-Hot Valley

  Sustainability · 2025-12-08

  articleOpen accessSenior author

  Conventional tourism planning in ecologically fragile regions often adopts a reductionist perspective, failing to address the synergistic spatial interactions between ecological conservation, resource utilization, and infrastructure. To bridge this gap, this study develops a multi-constraint synergistic assessment framework for the dry-hot valley of Lujiang Dam (LJD) in China. Grounded in the understanding of rural tourism as a complex adaptive system, the framework innovatively integrates the InVEST model, kernel density estimation, and cumulative cost-distance algorithms to identify Natural Spatial Suitability for Tourism Development (NSSTD). Key findings include (1) pronounced spatial heterogeneity in habitat quality, with high-quality zones in the west/southeast requiring strict conservation; (2) a “barbell-shaped” clustering of natural/cultural resources at the valley’s northern and southern extremities, highly congruent with ethnic settlements; and (3) a “concentric layered” accessibility pattern where 88.08% of resources are within a 90 min drive. Crucially, the spatial overlay analysis revealed that NSSTD (54.74 km2) emerges not from single high-value zones but from areas of synergy, such as those with medium habitat quality coupled with high resource endowment and accessibility. These results provide a scientifically robust, spatially explicit layer for China’s “Multi-plan Integration” territorial spatial planning. They enable differentiated strategies—channeling development to southern corridors, implementing niche tourism in northern “structural hole” villages, and enforcing conservation in western habitats—thereby offering a replicable methodology to balance ecological integrity with sustainable rural development.

  PublisherOA PDFDOI

• Circulating immune biomarkers correlating with response in patients with metastatic renal cell carcinoma on immunotherapy

  JCI Insight · 2025-01-07 · 4 citations

  articleOpen access

  Since multiple front-line immune checkpoint inhibitor-based (ICI-based) combinations are approved for metastatic renal cell carcinoma, biomarkers predicting for ICI responses are needed past clinical prognostication scores and transcriptome gene expression profiling. Circulating markers represent opportunities to assess baseline and dynamic changes in immune cell frequency and cytokine levels while on treatment. We conducted an exploratory prospective correlative study of 33 patients with metastatic clear cell renal cell carcinoma undergoing treatment with ICIs and correlated changes in circulating immune cell subsets and cytokines with clinical responses to treatment. Cell frequencies and cytokine levels were compared between responders and nonresponders using unpaired parametric t tests, using prespecified alpha level of significance of 0.05. Classical monocyte subsets (CD14+CD16-), as well as 7 cytokines (IL-12/23 p40, macrophage inflammatory protein-1a, macrophage inflammatory protein-1b, vascular cell adhesion molecule-1, intercellular adhesion molecule-1, IL-8, and TNF-α) were higher at baseline for responding versus nonresponding patients. Dynamic changes in thymus- and activation-regulated chemokine (TARC), placental growth factor (PlGF), and vascular endothelial growth factor (VEGF) also correlated with patients with ICI response. In summary, macrophage-activating agents were observed to be important in ICI response and may highlight the importance of the innate immune response in ICI responses.

  PublisherOA PDFDOI

• Sequential design for paired ordinal categorical outcome

  Statistical Methods in Medical Research · 2025-03-31

  articleSenior authorCorresponding

  This study addresses a critical gap in the design of clinical trials that use grouped sequential designs for one-sample or paired ordinal categorical outcomes. Single-arm experiments, such as those using the modified Rankin Scale in stroke trials, underscore the necessity of our work. We present a novel method for applying the Wilcoxon signed-rank test to grouped sequences in these contexts. Our approach provides a practical and theoretical framework for assessing treatment effects, detailing variance formulas and demonstrating the asymptotic normality of the U-statistic. Through simulation studies and real data analysis, we validate the empirical Type I error rates and power. Additionally, we include a comprehensive flowchart to guide researchers in determining the required sample size to achieve specified power levels while controlling Type I error rates, thereby enhancing the design process of sequential trials.

  PublisherDOI

• Ultrasound-activated and P-selectin-targeted liposomes overcome biofilm barriers for surgical site infections therapy

  Journal of Nanobiotechnology · 2025-11-26

  articleOpen access

  Pseudomonas aeruginosa-induced biofilm-associated surgical site infections (BSSIs) pose a dual therapeutic challenge: the dense extracellular polymeric substances (EPS) barriers hinder drug enrichment and penetration, whereas the immunosuppressive microenvironment within the biofilm impedes infection clearance, leading to persistent bacterial colonization and recurrence. This study developed an ultrasound-activated P-selectin-targeted liposome (SPCMPL) integrating natural sulfatide ligands (targeting P-selectin overexpressed in inflamed BSSI vasculature), sonosensitizer chlorin e6 (Ce6), meropenem prodrug, and perfluoropentane (PFP) to achieve efficient antibiotic delivery and BSSI treatment via breaking through the biofilm barriers and activating immunomodulation. SPCMPL employed ligand/receptor-mediated transcytosis for enrichment in BSSI lesions, where the PFP phase transition triggered by ultrasound disrupted the biofilm EPS structure. This process can both trigger the in-situ generation of reactive oxygen species (ROS) by Ce6 and loosen the EPS matrix. This degradation then facilitated meropenem release, allowing it to penetrate the biofilm more effectively and achieve antimicrobial concentrations throughout. Furthermore, the mass-produced ROS polarized macrophages to a pro-inflammatory M1 phenotype, thereby enhancing phagocytosis, remodeling the microenvironment, and inhibiting biofilm persistence. Ultrasound-triggered spatial control localized antibiotic release and immunomodulation to the infection site, optimizing local delivery while minimizing systemic toxicity and reducing the risk of systemic cytokine storms. The results demonstrated that the SPCMPL with ultrasound manipulation integrated biofilm disruption, targeted drug release, and immunomodulation to completely eradicate both planktonic and biofilm-embedded bacteria and effectively treat BSSI.

  PublisherOA PDFDOI

Frequent coauthors

• M. W. Ahmed

  University of Missouri–Kansas City

  157 shared

• V. Popov

  Duke University

  123 shared

• S.F. Mikhailov

  Duke University

  116 shared

• B. A. Perdue

  70 shared

• H. R. Weller

  Duke University

  65 shared

• R. E. Pywell

  University of Saskatchewan

  60 shared

• V.N. Litvinenko

  57 shared

• B. Norum

  Duke University

  57 shared