About

Huiman Xie Barnhart is the James B. Duke Distinguished Professor of Biostatistics & Bioinformatics at Duke University. He also serves as a Professor of Biostatistics & Bioinformatics, the Associate Chair for Faculty Mentorship and Development, and is a member of the Duke Clinical Research Institute. His academic and professional roles are based at Duke's Department of Biostatistics and Bioinformatics, located at 2424 Erwin Road, Durham, NC. His research focuses on biostatistics and bioinformatics, contributing to the advancement of these fields through his leadership and scholarly work.

Research topics

• Medicine
• Chemistry
• Internal medicine
• Traditional medicine
• Intensive care medicine
• Food science
• Gastroenterology

Selected publications

• Identification of Reduced ERAP2 Expression and a Novel HLA Allele as Components of a Risk Score for Susceptibility to Liver Injury Due to Amoxicillin-Clavulanate

  UNC Libraries · 2026-02-17

  articleOpen access
  PublisherDOI

• Liver Injury due to Intravenous Methylprednisolone in the Drug‐Induced Liver Injury Network

  Liver International · 2025-01-13 · 5 citations

  articleOpen access

  BACKGROUND AND AIMS: Short courses of intravenous (iv) methylprednisolone (MP) can cause drug induced liver injury (DILI). The aim of this study was to assess the clinical features and HLA associations of MP-related DILI enrolled in the US DILI Network (DILIN). METHODS: DILIN cases with MP as a suspected drug were reviewed. DILIN causality scoring was assigned on a 5-point scale (definite, highly likely, probable, possible, unlikely). All cases with MP causality scores of definite, highly likely or probable were analysed. HLA data from direct sequencing were analysed. RESULTS: Eleven cases of definite, highly likely, or probable MP DILI were identified. The median age was 48 years; 73% were female; median latency to onset was 30 days; 55% were jaundiced; and all had hepatocellular injury with one patient requiring transplantation. Nine of the 11 cases were in patients with multiple sclerosis (MS). Liver biopsies in 7 cases revealed mild acute hepatitis with/without cholestasis. HLA data demonstrated that HLA-DRB1*15:01, the primary HLA class II allele associated with MS was over-represented. HLA-DQB1*06:02-HLA-DQA1*01:02 which is haplotypic with the HLA-DRB1*15 haplotype was more common in the MP DILI cases compared to other DILI controls (p = 0.03) and to DILI controls exposed to MP (p = 0.04). CONCLUSION: MP DILI is characterised by hepatocellular injury, short latency and generally rapid recovery. There was no independent HLA haplotype associated with MP DILI.

  PublisherOA PDFDOI

• Associations of gender and a proxy of female menopausal status with histological features of drug‐induced liver injury

  UNC Libraries · 2025-09-13

  articleOpen access

  BACKGROUND & AIM: Gender and menopause may contribute to type and severity of drug-induced liver injury (DILI) by influencing host responses to injury. The aim of this study was to assess the associations of gender and female age 50 [a proxy of menopause] with histological features of liver injury in 212 adults enrolled in the Drug-Induced Liver Injury Network (DILIN) registry. METHODS: All participants had a causality score of at least 'probable', a liver biopsy within 30 days of DILI onset, and no prior chronic liver disease. Biochemical and histological injury types were classified as hepatocellular or cholestatic/mixed injury. The cohort was divided into three gender/age categories: men (41.0%), women <50 years (27.4%) and women ≥50 years of age (31.6%). Interaction of gender and age category (≥50 or not) was assessed. RESULTS: Hepatocellular injury was more prevalent in women <50 years vs. others (P=.002). After adjusting for biochemical injury types, black race and possible ageing effects, more severe interface hepatitis was noted in biopsies of women <50 years compared to those of men and women ≥50 years (P=.009 and P=.055 respectively). Compared to those of men, biopsies of women showed greater plasma cell infiltration, hepatocyte apoptosis, hepatocyte rosettes and lobular disarray but less iron-positive hepatocytes and histological cholestasis (P<.05). These associations persisted after excluding cases of amoxicillin/clavulanic acid, anabolic steroids or nitrofurantoin DILI which showed gender-specific distributions. CONCLUSION: Gender and a proxy of menopause were associated with various features of inflammation and injury in DILI.

  PublisherDOI

• DRESS Syndrome in Patients With Drug-Induced Liver Injury: Characteristics and HLA Risk Factors

  The American Journal of Gastroenterology · 2025-06-12 · 3 citations

  articleOpen access

  INTRODUCTION: Drug reaction with eosinophilia and systemic symptoms (DRESS) can sometimes occur in patients with drug-induced liver injury (DILI). However, detailed studies of DRESS in patients with DILI from the United States are lacking. We investigated the characteristics and human leukocyte antigen (HLA) risks for DILI who also developed DRESS. METHODS: Patients with definite, highly likely, or probable DILI enrolled into US DILI Network studies between September 2004 and August 2023 were included. DRESS was defined based on modified RegiSCAR criteria. HLA alleles were compared between DILI-DRESS cases and 2 control groups (DILI with non-DRESS rash [n = 244] and DILI without rash [n = 1,637]). RESULTS: Of 2,121 participants with DILI during the study period, 128 participants had DRESS (6%). The most frequently implicated drugs causing DRESS were trimethoprim/sulfamethoxazole, lamotrigine, phenytoin, allopurinol, and vancomycin. Compared with 1993 patients with DILI without DRESS, patients with DILI + DRESS were younger (mean age 42.3 years vs 50.6 years), were more likely to be Black (26% vs 12%), and had shorter latency (median 31 days vs 47 days), higher frequency of rash (100% vs 13%), eosinophilia (55% vs 13%), and fever (76% vs 16%) ( P < 0.001 for all). Compared with DILI without DRESS, DILI + DRESS had more severe liver injury (severe/fatal: 45% vs 21.5%, P < 0.001) and higher overall (15.6% vs 6.3%, P < 0.001) and liver-related (9% vs 2.3%, P < 0.001) mortality. HLA A*32:01 , HLA B*53:01 , and HLA B*58:01 were significantly enriched in DILI-DRESS cases, compared with control groups. DISCUSSION: Patients with DILI and DRESS are younger, are more likely to be Black, have shorter time to DILI onset with more severe liver injury and higher overall and liver-related mortality. HLA A*32:01 , HLA B*53:01 , and HLA B*58:01 are risk factors for DILI-DRESS.

  PublisherOA PDFDOI

• Clinical and HLA Associations of Fluoroquinolone-Induced Liver Injury: Results From the Drug-Induced Liver Injury Network

  The American Journal of Gastroenterology · 2025-04-10 · 6 citations

  articleOpen access

  INTRODUCTION: Fluoroquinolones (FQ) have a favorable safety profile, but the risk of drug-induced liver injury (DILI) is well described. The aim of this study was to identify clinical features and HLA genetic variants associated with FQ-DILI in a large national registry. METHODS: Analysis of FQ-DILI cases enrolled in DILI Network between 2004 and 2022. HLA class I and II alleles were sequenced by the Illumina MiSeq platform. RESULTS: Sixty-one cases (32 ciprofloxacin, 22 levofloxacin, 7 moxifloxacin) were included. Clinical features between the 3 drugs were similar. The median duration of therapy was 7 (range 2-54) days, median age 53 (range 22-80) years, and 67% were female. Median latency to onset was 12 (range 2-1,370) days with 44% hepatocellular, 30% mixed, and 26% cholestatic pattern of liver injury. Median time to recovery was 65 days, but 13% had persistent injury at 6 months, 15% died (11% because of liver failure). Two HLA alleles were associated with an increased risk of liver injury: HLA-DQA1*03:01 (carriage frequency 38% in cases vs 19% in controls) and HLA B*57:01 (15% vs 6%). There was a significant difference between the combined carriage frequency of the 2 alleles of 48% in cases vs 24% controls ( P = 0.0001). No clinical characteristics or outcomes were associated with carriers compared with noncarriers. DISCUSSION: FQ DILI is a class effect that presents with a short latency, variable pattern of liver injury, and carries a significant risk of chronicity and mortality. There is a significant association with HLA-DQA1*03:01 and HLA B*57:01 .

  PublisherOA PDFDOI

• Refinement of Hy's Law using the Drug-Induced Liver Injury Network Database.

  UNC Libraries · 2025-10-19

  articleOpen access

  BACKGROUND: Hyman Zimmerman observed that hepatocellular (HC) drug-induced liver injury (DILI) with jaundice had a mortality rate of ≥10% (Hy's Law). Hy's Law does not specify the timing of liver tests nor the definition of HC DILI versus cholestatic or mixed (C/M) DILI. We aimed to assess the validity of Hy's Law in the prospective Drug-Induced Liver Injury Network (DILIN) cohort. METHODS: Drugs with ≥10 confirmed DILI cases with jaundice were analyzed. Four permutations of Hy's Law were applied: R≥ 5 using initial (1) or peak (2) ALT, AST and Alk P levels, and the FDA associated criteria of ALT or AST ≥ 3x ULN with Alk P ≤ 2x ULN using initial (3) or peak values (4). Mortality was death or liver transplant adjudicated to be due to DILI. RESULTS: Using initial R values, mortality was 11.1% for HC vs 2.0% for C/M (p<0.001); using peak R values, mortality was 10.3% vs 1.6% (p<0.001). Using FDA associated definition, mortality was 7.9% vs 3.9% (p=0.04) using initial values and 7.9% vs 3.0% (p=0.01) using peak values. Using initial R values, drugs that frequently caused HC injury generally had mortality rates ≥ 10%; while drugs that typically caused C/M injury all had rates <10%. Occasional agents that caused HC injury with jaundice were associated with low mortality. CONCLUSIONS: Initial R values were the most reliable means of identifying Hy's Law cases. There were some drugs that caused HC injury with jaundice but with mortality rates <10%. Refinement of Hy's Law is warranted.

  PublisherDOI

• FRI-153 Prognostic significance of very high and towering peak alanine aminotransferase (ALT) values with normal or near normal total bilirubin in patients with idiosyncratic drug-induced liver injury

  Journal of Hepatology · 2025-05-01

  article
  PublisherDOI

• THU-352 The histological activity index is associated with disease severity and treatment response in patients with immune mediated liver injury from checkpoint inhibitors

  Journal of Hepatology · 2025-05-01

  article
  PublisherDOI

• Impact of Individualized Versus Weight-Based Pain Protocols on Patient Satisfaction for Patients With Sickle Cell Disease Experiencing a Vaso-Occlusive Episode

  Journal of Emergency Nursing · 2025-02-26

  articleOpen access
  PublisherOA PDFDOI

• Sample Size and Power Calculations With Win Measures Based on Hierarchical Endpoints

  Statistics in Medicine · 2025-05-01 · 3 citations

  article1st authorCorresponding

  Win measures, such as win ratio, win odds, net benefit, and desirability of outcome ranking (DOOR), have become popular approaches for the analysis of hierarchical endpoints in clinical studies. Sample size and power calculations with win measures based on hierarchical endpoints are often based on simulation studies that can be cumbersome. Existing sample size and power formulas require investigators to specify clinically significant and meaningful magnitudes of win measures and probability of ties that are difficult to elicit based on prior published literature or preliminary data. In this paper, we provide sample size and power calculation formulas for the four win measures. To facilitate the formula-based sample size or power calculations, we provide formulas to compute overall win measures and overall probability of ties needed by using the specification of marginal win measures and marginal probability of ties that are readily available from clinical investigators or literature. The latter formulas provide a novel way to specify a meaningful and justifiable magnitude of win measures and the magnitude of probability of ties. Therefore, they can be readily used to evaluate the powers based on the number of multiple endpoints, the ordering, and types of endpoints. Our extensive simulation studies show that the power estimations based on these formulas are often like the simulated powers for any type of correlated hierarchical endpoints except for scenarios with very high correlations between endpoints. We illustrate the usefulness of our formulas by using data from three trials with different types of hierarchical endpoints.

  PublisherDOI

Recent grants

• NIH Grant R01DK080094

  NIH · $485k · 2012

• Drug Induced Liver Injury Network Data Coordinating Center

  NIH · $23.8M · 2003–2028

• NIH Grant R29HL058014

  NIH · $541k · 2003

• Coordinating Center for the Drug Induced Liver Injury Network (DILIN)

  NIH · $7.6M · 2003–2023

Frequent coauthors

• Evan R. Myers

  Duke University

  335 shared

• Nicholas A. Cataldo

  314 shared

• Peter G. McGovern

  311 shared

• Sandra Ann Carson

  Yale University

  310 shared

• William D. Schlaff

  Thomas Jefferson University

  309 shared

• Michael P. Steinkampf

  308 shared

• Michael P. Diamond

  308 shared

• John E. Nestler

  European Gravitational Observatory

  308 shared

Education

• PhD, Statistics

  University of Pittsburgh

  1992