Abstract: BACKGROUND: Knee osteoarthritis (KOA) is a common chronic inflammatory disease that causes damage to joint cartilage and surrounding tissues. Immune cells play an important role in the immune-inflammatory response in knee osteoarthritis, but the specific mechanisms involved are still not fully understood. 
OBJECTIVE: To evaluate the potential causal relationship between 731 immune cell phenotypes and the risk of knee osteoarthritis using Mendelian randomization. 
METHODS: Summary statistics of genome-wide association studies (GWAS) for 731 immune cell phenotypes (from GCST0001391 to GCST0002121) obtained from the GWAS catalog and GWAS data for knee osteoarthritis from the IEUGWAS database (ebi-a-GCST007090) were used. Inverse variance-weighted method, MR-Egger regression, weighted median method, weighted mode method, and simple mode method were employed to investigate the causal relationship between immune cells and knee osteoarthritis. Sensitivity analyses were conducted to assess the reliability of the Mendelian randomization results. Reverse Mendelian randomization analysis was also performed using the same methods. 
RESULTS AND CONCLUSION: The forward MR analysis indicated significant causal relationships (FDR < 0.20) between knee osteoarthritis and four immune cell phenotypes, namely CD27 on CD24+CD27+ in B cells (OR=1.026, P=0.000 26, Pfdr=0.18), CD33 on CD33dim HLA DR- in myeloid cells (OR=1.014, P=0.000 50, 
Pfdr=0.18), and CD45RA+CD28-CD8br %CD8br in Treg cells (OR=1.001, P=0.000 78, Pfdr=0.18), and PDL-1 on monocytes in mononuclear cells (OR=0.952, P=0.000 98, Pfdr=0.18). These immune cell phenotypes showed direct positive or negative causal associations with the risk of knee osteoarthritis. Reverse Mendelian randomization analysis revealed no significant causal relationships (FDR < 0.20) between knee osteoarthritis as exposure and any of the 731 immune cell phenotypes. The results of sensitivity analysis show that the P-values of the Cochran’s Q test and the MR-Egger regression method for bidirectional Mendelian randomization were both greater than 0.05, indicating that there is no significant heterogeneity and pleiotropy in the causal effect analysis between immune cell phenotypes and knee osteoarthritis. To conclude, there may be four potential causal relationships between immune cell phenotypes, such as CD27 on CD24+CD27+ cells, CD33 on CD33dim HLA DR- cells, CD45RA+CD28-CD8br %CD8br cells, and PDL-1 on monocytes, and knee osteoarthritis. These findings provide valuable clues for studying the biological mechanisms of knee osteoarthritis and exploring early prevention and treatment strategies. They also offer new directions for the development of intervention drugs.

Key words: osteoarthritis, knee osteoarthritis, immune mechanism, immune cell, Mendelian randomization, causal relationship, genome-wide association analysis, inverse variance-weighted method, sensitivity analysis