UDCA May Promote COVID-19 Recovery: A Cohort Study with AI-Aided Analysis

Yang Yu, Guo Yu, Luyao Han, Jian Li, Zhilong Zhang, Tian-Shuo Liu, Mingfeng Li, De‐Chuan Zhan, Shaoqiu Tang, Zhi-Hua Zhou, Guangji WangmedRxiv 2023 · DOI

Abstract

To investigate the impact of ursodeoxycholic acid (UDCA) treatment on the clinical outcome of mild and moderate COVID-19 cases, a retrospective analysis was conducted to evaluate the efficacy of UDCA on patients diagnosed with COVID-19 during the peak of the Omicron outbreak in China. This study presents promising results, demonstrating that UDCA significantly reduced the time to Body Temperature Recovery after admission and a higher daily dose seems to be associated with a better outcome without observed safety concerns. We also introduced VirtualBody, a physiologically plausible artificial neural network model, to generate an accurate depiction of the drug concentration-time curve individually, which represented the absorption, distribution, metabolism, and excretion of UDCA in each patient. It exhibits exceptional performance in modeling the complex PK-PD profile of UDCA, characterized by its endogenous and enterohepatic cycling properties, and further validates the effectiveness of UDCA as a treatment option from the drug exposure-response perspective. Our work highlights the potential of UDCA as a novel treatment option for periodic outbreaks of COVID-19 and introduces a new paradigm for PK-PD analysis in retrospective studies to provide evidence for optimal dosing strategies.

Introduction

While oral antivirals for COVID-19 have become available, high costs and limited accessibility remain significant barriers globally. Researchers from Nanjing University and China Pharmaceutical University have explored a potential alternative: ursodeoxycholic acid (UDCA). UDCA is a widely available liver medication known to inhibit the FXR receptor, which regulates the expression of ACE2—the primary entry point for SARS-CoV-2. This study evaluates UDCA's clinical efficacy using a retrospective cohort and a novel AI-driven pharmacokinetic framework.

Methods

The study utilized a dual-methodology approach to assess UDCA's impact on 115 non-severe COVID-19 patients (65 treated, 50 control) during the 2022 Omicron surge in China.

Clinical Retrospective Cohort

The primary endpoint was "Time to Body Temperature Recovery." Statistical analyses compared standard care against UDCA treatment, adjusting for various clinical covariates.

VirtualBody AI Framework

To address the lack of longitudinal blood-concentration data typical of retrospective studies, the authors developed VirtualBody. This physiologically structured neural network models the absorption, distribution, metabolism, and excretion (ADME) of UDCA. Unlike standard deep learning models, VirtualBody incorporates a "Skeleton Model" to simulate drug transitions between organs, accounting for UDCA's complex enterohepatic circulation.

Key Results

Accelerated Clinical Recovery

The clinical analysis revealed that UDCA treatment significantly shortened the duration of fever. Patients in the UDCA group achieved a median body temperature recovery in 4 days, compared to 5 days in the control group ($P = 0.030$).

Cumulative recovery of body temperature in UDCA-treated vs. control groups.

Dose-Response Relationship

The study further investigated the impact of dosage on recovery. Analysis showed that the therapeutic benefit was dose-dependent. Patients receiving a daily dose of 300mg or more experienced significantly faster recovery (Hazard Ratio 1.82), whereas the 150mg dose did not yield a statistically significant improvement over the control group.

Cumulative recovery stratified by UDCA dosage (150mg vs. 300mg+).

AI Model Performance and Precision Dosing

VirtualBody significantly outperformed traditional pharmacokinetic models (like PopPK) and standard AI architectures (LSTMs). By calculating the individual Area Under the Curve (AUC), the researchers established a clear exposure-response relationship. Further analysis using decision trees on a virtual population of 1,000 individuals identified age and liver function (AST and Albumin) as key predictors of drug exposure, suggesting that dosing should be tailored to individual physiological profiles.

Discussion and Limitations

The findings suggest that UDCA is a safe and effective candidate for accelerating COVID-19 recovery. However, the authors note several limitations, including the small sample size (n=115) and the retrospective nature of the study. The cohort primarily consisted of young, vaccinated individuals, meaning results may not generalize to elderly or high-risk populations. Future large-scale prospective randomized controlled trials are necessary to confirm these results and further refine UDCA's role in COVID-19 management.

Summary updated

Abstract

Abstract To investigate the impact of ursodeoxycholic acid (UDCA) treatment on the clinical outcome of mild and moderate COVID-19 cases, a retrospective analysis was conducted to evaluate the efficacy of UDCA on patients diagnosed with COVID-19 during the peak of the Omicron outbreak in China. This study presents promising results, demonstrating that UDCA significantly reduced the time to Body Temperature Recovery after admission and a higher daily dose seems to be associated with a better outcome without observed safety concerns. We also introduced VirtualBody, a physiologically plausible artificial neural network model, to generate an accurate depiction of the drug concentration-time curve individually, which represented the absorption, distribution, metabolism, and excretion of UDCA in each patient. It exhibits exceptional performance in modeling the complex PK-PD profile of UDCA, characterized by its endogenous and enterohepatic cycling properties, and further validates the effectiveness of UDCA as a treatment option from the drug exposure-response perspective. Our work highlights the potential of UDCA as a novel treatment option for periodic outbreaks of COVID-19 and introduces a new paradigm for PK-PD analysis in retrospective studies to provide evidence for optimal dosing strategies.
BibTeX
@misc{yu2023udca,
  author = {Yang Yu and Guo Yu and Luyao Han and Jian Li and Zhilong Zhang and Tian-Shuo Liu and Mingfeng Li and De‐Chuan Zhan and Shaoqiu Tang and Zhi-Hua Zhou and Guangji Wang},
  title = {{UDCA May Promote COVID-19 Recovery: A Cohort Study with AI-Aided Analysis}},
  year = {2023},
  doi = {10.1101/2023.05.02.23289410},
  howpublished = {medRxiv},
}
Figure