Letter to the Editor: Is Omeprazole a contributing risk factor for osteoarthritis (OA)?

Haoyang Zhang, Yong‐Ping Chen, Liangbin Zhou, Liming ZhengJournal of Translational Medicine 2024 · DOI · OpenAlex

Abstract

Recent research suggested a causal link between omeprazole use and the development of osteoarthritis (OA) using Mendelian Randomization (MR). This paper provides a critical re-evaluation of those findings, identifying significant methodological flaws including genetic pleiotropy and sample overlap. By applying more rigorous MR techniques and independent validation datasets, the authors demonstrate that the purported causal relationship is statistically non-significant, refuting the claim that omeprazole is a risk factor for OA.

Introduction

Mendelian Randomization (MR) serves as a robust statistical framework, utilizing genetic variants as instrumental variables to infer causal relationships between exposures and outcomes. A recent study by Cao et al. applied MR to conclude that omeprazole, a widely used proton-pump inhibitor (PPI), contributes to osteoarthritis. Given the global prevalence of omeprazole prescriptions, such a claim carries significant clinical weight. However, methodological "red flags"—specifically violations of core MR assumptions—prompted a re-analysis to determine the validity of these findings.

Methods

The authors re-evaluated the original data using refined MR techniques to address potential biases:

  • Sensitivity Analysis for Pleiotropy: Genetic variants (SNPs) were screened for pleiotropy. Three specific SNPs were identified as influencing OA through pathways independent of omeprazole and were subsequently removed.
  • Two-Sample MR Validation: To avoid "one-sample" bias, where exposure and outcome data come from the same population (UK Biobank), the authors validated the exposure against the arcOGEN study, an independent European osteoarthritis cohort.
  • Standard MR Estimators: Consistency was checked using Inverse Variance Weighted (IVW), MR-Egger, Weighted Median, and Weighted Mode methods.

Results

The re-analysis fundamentally shifted the conclusions of the original study. Upon removing pleiotropic SNPs, the IVW fitting slopes decreased, rendering the relationship between omeprazole and OA statistically non-significant.

Crucially, when the independent arcOGEN dataset was employed to eliminate sample overlapping bias, the results across all 12 distinct analyses were universally non-significant. This lack of statistical evidence directly refutes the previous assertion of causality.

Figure 6: Forest plot demonstrating universally non-significant causal effects (β) of omeprazole on OA across multiple MR methods in the independent validation dataset.

Discussion

The discrepancy between the original study and this re-analysis highlights the "one-sample trap." Using overlapping populations for both exposure and outcome in a two-sample MR design can generate false positives. Furthermore, the biological complexity of OA, involving distinct roles for osteoclasts at different disease stages, suggests that a simple binary classification in genetic studies may overlook critical nuances.

Conclusion

The evidence supporting omeprazole as a causal risk factor for osteoarthritis is weak and fails to withstand rigorous methodological scrutiny. Future MR studies in this domain should prioritize non-overlapping datasets and stringent filters for pleiotropic variants to ensure clinical recommendations are based on sound causal inference.

Summary updated

BibTeX
@article{zhang2024letter,
  author = {Haoyang Zhang and Yong‐Ping Chen and Liangbin Zhou and Liming Zheng},
  title = {{Letter to the Editor: Is Omeprazole a contributing risk factor for osteoarthritis (OA)?}},
  journal = {Journal of Translational Medicine},
  year = {2024},
  doi = {10.1186/s12967-024-05525-9},
}
Figure