2565 Background: Mesothelin (MSLN) is overexpressed in mesothelioma and other solid tumors making it the focus of multiple targeted therapies, including antibody drug conjugates (ADCs), T cell receptor (TCR) fusion constructs, and chimeric antigen receptor T-cell (CAR-T) products. However, the clinical efficacy of these treatments remains limited. We hypothesize that soluble MSLN (sMSLN) in the plasma binds to MSLN-targeted therapies before reaching the tumor. Our study aimed to evaluate the effects of sMSLN on a MSLN-targeting antibody, anetumab, and explore methods to reduce sMSLN. Methods: Exploratory analysis in NCT03126630: sMSLN testing was performed on blood samples from participants enrolled in NCT03126630 at the Molecular Targets Core Lab, National Cancer Institute. The median was used to categorize high and low levels of sMSLN. Survival was estimated with the Kaplan-Meier method and compared between groups with the log rank test. Anetumab immunoprecipitation: Anetumab was covalently coupled to Dynabeads at 5 µg per mg of beads to immunoprecipitate MSLN from two plasma samples. Assays were performed in duplicate. Therapeutics plasma exchange (TPE):Whole blood samples were collected before and after one plasma volume of TPE with albumin as the replacement fluid in patients undergoing routine TPE for various medical conditions, such as autoimmune diseases and hyperviscosity syndromes. Plasma levels of sMSLN were measured with an ELISA assay in matched pre- and post-TPE plasma samples. Results: We obtained sMSLN levels from 40 patients enrolled in NCT03126630. For patients treated with both anetumab ravtansine and pembrolizumab, median progression free survival (PFS) was shorter in the high sMSLN group (5 months) vs the low sMSLN group (12 months). For patients treated with pembrolizumab alone, PFS was similar for patients with high and low sMSLN (4 vs 3.4 months). Anetumab significantly reduced the concentration of sMSLN in plasma samples as detected by MSLN ELISA (p<0.05), demonstrating that sMSLN can bind to and sequester anetumab. Next, we evaluated TPE as a mechanism to reduce sMSLN. We obtained pre- and post-TPE plasma samples from 15 patients undergoing routine TPE for various medical conditions. TPE consistently reduced sMSLN (p<0.05) with an average decrease of 43.6% or 15.4 ng/mL. Conclusions: We found that high sMSLN levels are associated with shorter PFS for anetumab ravtansine, but not pembrolizumab. Additionally, anetumab binds to sMSLN in the plasma, which suggests that sMSLN can sequester anti-MSLN antibodies and may limit the efficacy of MSLN-targeted therapies. High levels of sMSLN could potentially be used as a biomarker to select which patients should not receive MSLN-targeting therapies. Furthermore, our results indicate that sMSLN reduction is feasible with TPE. Decreasing sMSLN could restore the activity of MSLN-targeted therapies.