Abstract
With the rising incidence of rheumatoid arthritis (RA) and the increasing percentage of serum RF negativity, more and more accurate methods are urgently needed for the early diagnosis and prevention of RA, among which serum albumin (ALB) is closely related to the development of RA, and it is expected to become a new auxiliary diagnostic means, but its relationship with RA is not clear, so the present study aimed to investigate the Causal relationship. In this study, we used a generalized linear model and smoothed curve fitting to comprehensively evaluate the relationship between ALB and RA through the data of ALB and RA in the NHANES database, in addition, we further used inverse variance weighted (IVW) in Mendelian randomization (MR) in conjunction with the other 4 methods to further validate and clarify the causal relationship. The results were also examined for heterogeneity and horizontal pleiotropy to assess whether the results were robust. Finally, we used Bayesian co-localization analysis to clarify that ALB and RA share common genetic loci. In the observational study, after correction for multiple confounders, ALB remained more significantly negatively associated with RA (OR = 0.66, [95% CI = 0.51-0.86], P = .003), and subgroup analyses showed significant negative associations in both men and women (men: OR = 0.67, [95% CI = 0.46-0.99], P = .046; females: OR = 0.66, [95% CI = 0.44-1.00], P = .049). In further MR analysis, IVW: ALB on RA, OR = 0.70 [95% 0.52-0.93], P = .013; RA on ALB, OR = 0.95 [95% CI = 0.93-0.98], P < .001. The results of the MR analyses were in agreement with those of NHANES, which did not share a common genetic locus in co-localization analysis. There is a significant relationship between ALB and RA, and the reduction of ALB may be one of the risk factors for RA, as well as one of the outcomes in the development of RA.