Multivariable Mendelian Randomization adjusting for heritable confounding analyzes the causal effects of C-reactive protein on multiple diseases

Background

C-reactive protein (CRP) is a marker of inflammation associated with autoimmune, cardiovascular, and neuropsychiatric disorders. However, it remains unclear whether CRP causally affects these traits or if observed associations result from reverse causation or confounding. Mendelian randomization (MR) uses genetic variants as instrumental variables to estimate causal effects and avoid the biases present in observational studies. Prior MR studies have suggested causal effects of CRP on several traits, including schizophrenia, bipolar disorder, and colorectal cancer. However, MR may produce biased results if factors that confound the exposure and outcome are heritable, resulting in horizontal pleiotropy. This is a major concern for studies of CRP, because CRP levels may increase in response to inflammation caused by a wide range of heritable conditions.

Methods

Multivariable Mendelian randomization (MVMR) can be used to eliminate bias from heritable confounding when GWAS summary data are available for confounders. In this study, we use MVMR to estimate the causal effects of CRP on 12 outcomes with prior evidence of a causal or associational link to CRP. We use a novel computational pipeline to identify a broad set of potential heritable confounders between CRP and each outcome trait from studies in the MRC-IEU OpenGWAS database. We compare MVMR results with computationally selected confounders to univariable MR results and MVMR using a narrower, literature derived set of confounders.

Results

We find that univariable MR suggests evidence of a potential risk-increasing effect of CRP on coronary artery disease, knee osteoarthritis, and rheumatoid arthritis, and a protective effect on schizophrenia. However, after adjusting for computationally selected heritable confounders, only the causal effects on rheumatoid arthritis (OR 1.18, 95% CI [1.07,1.31], p=0.0010 by GRAPPLE) and schizophrenia (OR 0.87, 95% CI [0.79,0.96], p=0.0038 by GRAPPLE) remain significant. Additionally, after adjusting for confounders we find evidence of a potential protective effect of CRP on bipolar disorder at the nominal significance level, which is not observed in the univariable analysis.

Conclusion

These results suggest that univariable MR analyses of CRP may be biased by high levels of heritable confounding, though CRP may indeed play a causal role in development of some diseases, potentially mediated by its role in innate immunity. These results also high-light the potential for automatic confounder selection to improve the robustness of Mendelian randomization analyses.

Key Messages

• We used multivariable Mendelian randomization to estimate the causal effect of CRP levels on various diseases after adjusting for heritable confounders.

• We proposed a novel computational pipeline for phenome-wide heritable confounder selection using traits from the MRC-IEU OpenGWAS database.

• Our study did not find evidence of a causal effect of CRP on multiple diseases, except rheuma-toid arthritis and schizophrenia.

• Our study highlights the role of CRP as an indicator of inflammation rather than a causal factor in disease risk, suggesting that previous MR analyses may have been biased by heritable confounding.