Serum Biochemical Markers and Rheumatoid Arthritis: Insights From Observational Study and Mendelian Randomization Analysis

Rheumatoid arthritis (RA) is a chronic autoimmune disease that primarily affects the joints and is characterized by synovitis, resulting in pain, physical disability, and limited mobility. Serum albumin and globulin have been identified as potential biomarkers for assessing the inflammatory aging process in individuals with RA, reflecting their utility in monitoring disease progression and therapeutic responses [1, 2]. Emerging evidence suggests that systemic inflammation and nutritional status are closely associated with RA pathogenesis and progression [3-5]. However, which serum biochemical markers have a causal association with RA and can serve as potential auxiliary diagnostics for RA remains unknown. Currently, the combination of epidemiological research and Mendelian randomization (MR) for causal inference has become a popular analytical approach. Our study employed cross-sectional data from the National Health and Nutrition Examination Survey (NHANES), which provides a valuable resource for studying the relationship between serum biochemical markers and RA. By combining MR and Generalized Summary-data-based MR (GSMR) analyses, we employ genetic variation as an instrumental variable to infer causal relationships. This study aims to identify specific serum biochemical markers that are causally associated with RA risk, as shown in Figure 1. This cross-sectional study used data from a total of 10 cycles from the NHANES database from 2001 to 2020. Participants with complete demographic data and both serum biochemical markers and RA data were included in the study. A total of 5160 participants were included (Figure S1). Serum biochemical marker data were obtained from the BIOPRO (Standard Biochemistry Profile) project. The panel included 22 biomarkers: albumin, BUN (blood urea nitrogen), calcium, cholesterol, creatinine, globulin, glucose, iron, LDH (lactate dehydrogenase), phosphorus, total bilirubin, total protein, triglycerides, uric acid, AST (aspartate aminotransferase), ALT (alanine aminotransferase), HCO3 (bicarbonate), chloride, gamma GT (gamma-glutamyl transferase), potassium, sodium, and osmolality. This MR study employed two-sample MR [6] and GSMR analysis methods (Figure S2) [7]. The exposure factors comprised serum biochemical markers that exhibited significant associations with RA. Genome-wide association study (GWAS) data for exposure were obtained from previous research conducted by Nasa et al. comprised 363 228 individuals. GWAS data for outcomes were derived from a large-scale genetic meta-analysis of RA, including 14 361 cases and 43 923 controls of European descent. For the replication analysis, RA was selected as the outcome from FinnGen R10. STROBE-MR was employed as the reporting standard (Table S1). In the epidemiological research from the NHANES database, after integrating other covariates for analytical purposes, a final cohort of 5160 participants was available for analysis, including 1416 individuals with RA and 3744 without RA (Table S3). The survey-weighted multiple logistic regression model revealed that albumin, calcium, globulin, and uric acid remained significantly associated with the prevalence of RA after adjusting for all covariates (Figure 2A). Further, using the four-node restricted cubic spline (RCS) model (Tables S4–S8), our analysis showed non-linear U-shaped dose–response relationships between albumin, uric acid and the prevalence of RA (p of overall < 0.001, p of non-linear < 0.05), while calcium, globulin, and LDH exhibited linear relationships (all p nonlinear > 0.05, Figure 2B). Notably, within the physiological range, when albumin > 37 g/L, OR < 1 was a protective factor for RA (p = 0.03). GWAS data were collected (Table S2) and utilized for subsequent GSMR analysis, MR analysis, and meta-analysis. However, due to the lack of GWAS data relating to globulin, analysis was limited to albumin, calcium, LDH, uric acid, and the risk of RA. The results of the inverse variance weighted (IVW) method showed that albumin remained significantly associated with the risk of RA (OR = 0.79, 95% CI: 0.67–0.92, p = 3.3e-03) (Tables S9 and S10, Figure S3). In contrast, multiple MR methods did not show any significant causal association between calcium, LDH, and urate with RA risk (Figure 2C). To account for biases in GWAS measurements across different databases and reinforce the robustness of the findings, RA data from FinnGen R10 were included for replication analysis, and an IVW meta-analysis method was employed. The IVW-meta-analysis results supported the findings presented in this paper (p = 0.002) and further indicated that albumin had a negative causal effect on the risk of RA (Figure 2D and Tables S11 and S12). Previous studies have demonstrated that albumin plays an important role in maintaining homeostasis and regulating the immune system, and this may be a mechanism by which albumin reduces the risk of RA [8, 9]. Our investigation, leveraging NHANES data, uncovered a negative correlation between albumin levels and RA risk, with a significant protective effect observed when albumin levels exceeded 37 g/L (OR < 1, p < 0.05). This finding suggests that higher albumin levels within the physiological range may reduce the risk of RA. Meanwhile, MR meta-analysis further reinforced the notion that reduced albumin levels may be an underlying cause of RA. These observations suggest that albumin holds promise as a protective factor against RA, implying the feasibility of albumin supplementation as an effective therapeutic intervention to mitigate disease progression. High levels of uric acid have been shown to exacerbate inflammation, aggravating symptoms and driving disease progression of autoimmune inflammatory diseases, including RA [10, 11]. Hyperuricemia often coexists with metabolic diseases, such as hypertension, diabetes, and dyslipidemia, which may increase the risk of RA by activating or exacerbating inflammatory responses [12, 13]. In our study, survey-weighted multivariable regression analysis revealed individuals with high or low levels of uric acid were significantly more likely to develop RA compared to those with intermediate levels. Although LDH is not directly used as an indicator of inflammatory activity, high levels of LDH may indicate more severe inflammatory reactions or tissue damage in patients [14, 15]. It is plausible to suggest that high LDH levels may similarly indicate poor prognosis in RA patients. However, the robustness of these inferences necessitates rigorous validation through well-designed clinical trials to ensure their clinical reliability. The included genetic studies predominantly consist of individuals of European ancestry, limiting the generalizability. In conclusion, our study showed a linear relationship between calcium, globulin, LDH, and the risk of RA, while albumin and uric acid exhibited a U-shaped nonlinear association with RA. Further, MR analysis showed that albumin could be a protective factor in the development of RA. Thus, effective RA management strategies should integrate albumin level assessments, with particular attention to maintaining albumin levels above 37 g/L, alongside a focus on multiple risk factors. Further research is essential to explore the specific mechanisms in reducing RA risk. F.F. and H.T. mainly designed and performed analysis, verifying data, and wrote the manuscript; Y.X., C.Z, H.Q., H.W., W.Y., and A.Z. performed experiments and analysis; L.L. and C.L. supervised the entire project. All authors read, provided critical feedback on intellectual content, and approved the final manuscript. The authors thanked all participants in the study. The authors declare no conflicts of interest. The datasets analyzed in this study are publicly available summary statistics. GWAS summary statistics of serum albumin, calcium, globulin, LDH, and uric acid are publicly available from https://doi.org/10.35092/yhjc.12355382 and IEU OpenGWAS (https://gwas.mrcieu.ac.uk/). GWAS summary statistics of rheumatoid arthritis are publicly available from the EBI dataset (https://www.ebi.ac.uk/gwas/search?query) and the FinnGen consortium (https://www.finngen.fi/en/access_results). We used the GWAS Catalog web-based resource (https://www.ebi.ac.uk/gwas/). Figure S1: Workflow of NHANES. BIOPRO, standard biochemistry profile; BMI, body mass index; DM, diabetes. Figure S2: Workflow of Mendelian randomization. The MR approach was based on three key assumptions: (1) a strong correlation exists between genetic variants and exposure, (2) genetic variants are independent of confounding factors that influence the association between exposure and outcome, and (3) genetic variants solely impact the outcome through the specific exposure of interest. SNPs, single nucleotide polymorphisms. Figure S3: Leave-one-out plot and scatter plot between albumin and rheumatoid arthritis risk. MR, Mendelian randomization; SNPs, single nucleotide polymorphisms. Table S1: STROBE-MR checklist of recommended items to address in reports of Mendelian randomization studies. Table S2: Mendelian randomization dataset source. Table S3: Basic characteristics of participants by rheumatoid arthritis in adults, NHANES 2001–2020. Table S4: Association of Albumin with Rheumatoid arthritis, NHANES, 2001–2020. Table S5: Association of Calcium with Rheumatoid arthritis, NHANES, 2001–2020. Table S6: Association of Globulin with Rheumatoid arthritis, NHANES, 2001–2020. Table S7: Association of Lactate Dehydrogenase with Rheumatoid arthritis, NHANES, 2001–2020. Table S8: Association of Uric acid with Rheumatoid arthritis, NHANES, 2001–2020. Table S9: Instruments variables for albumin, calcium, LDH and uric acid. Table S10: GSMR and MR analysis results of Rheumatoid arthritis in the discovery and replication dataset. Table S11: Sensitivity analysis results of Rheumatoid arthritis in the discovery dataset. Table S12: Sensitivity analysis results of Rheumatoid arthritis in the replication dataset. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.