Abstract
BACKGROUND: Current evidence linking sedentary behavior (SB), physical activity (PA), and inflammation raises questions about their causal relationships, prompting concerns about potential residual confounding or reverse causation. METHODS: A bidirectional Mendelian randomization (MR) analysis was conducted. SB data (n = 408,815) from "computer use," "television watching," and "driving" were included. The PA data encompassed nine types of PA (n = 460,376) over the last four weeks and included data on the frequency of vigorous PA (n = 440,512) and moderate PA (n = 440,266) for over 10 min. Additionally, three genome-wide association study datasets (n = 64,949) on light, moderate, and vigorous exercise were included to minimize potential bias from changes in exercise intensity. Inflammation data included levels of C-reactive protein (CRP) (n = 575,531), glycoprotein acetyl (GlycA) (n = 115,082), interleukin (IL)-8, IL-6, IL-6 receptor (IL-6R), and soluble IL-6R (sIL-6R) (n = 35,278). All datasets represented participants of European ancestry. RESULTS: Television watching as an SB showed significant positive causal effects on GlycA and CRP (inverse variance weighted (IVW), odds ratios (OR): 1.34, 95% confidence intervals (CI): 1.25-1.44, p = 3.570 × 10-17; IVW, OR: 1.21, 95% CI: 1.16-1.26, p = 1.500 × 10-19, respectively), with more robust evidence for GlycA. In the direction from inflammation to PA, a negative causal relationship between CRP and"number of days/week of moderate PA 10+ minutes"was observed (IVW, OR: 0.92, 95% CI: 0.89-0.96, p = 3.260 × 10-5). Sensitivity analyses were used to verify the robustness and reliability of the results. However, other initially observed associations ceased to be significant after controlling for obesity-related confounders. CONCLUSION: Our MR analysis suggested a potential causal relationship between television watching and chronic low-grade inflammation, with more substantial evidence for GlycA. Additionally, different types of SB may have varying effects on inflammation. Obesity-related traits could partly or entirely influence the relationship between SB, PA, and inflammatory markers. Furthermore, Our findings indicate that SB is an independent risk factor for inflammation, separate from PA, and highlight the different mechanisms by which SB and PA affect disease.