#1044 Quantitative analysis of dialytic protein loss with medium cut-off and high-flux membranes

Abstract

Abstract Background and Aims Hemodialysis is an effective way for eliminating uremic retention products in patients with end-stage kidney disease. Medium cut-off membranes (MCO) were designed to enhance the dialytic removal of middle-sized molecules, such as beta-2 microglobulin (β2M). However, alongside these toxins, salutary proteins could also be lost. To better understand the spectrum of proteins cleared by MCO and high-flux (HF) dialyzers, we conducted qualitative and quantitative analysis of proteins in hemodialysis ultrafiltrate (UF). Method We established an ex vivo hemodialysis simulation system by using two Fresenius 2008T machines to concurrently dialyze a 2.5-liter human plasma reservoir for simultaneous testing of two dialyzers. We compared three dialyzers, including two HF dialyzers (Fresenius Medical Care Optiflux F180NR and FX CorAL 80), and an MCO membrane (Baxter Theranova 400; TH) (Fig. 1A). The blood flow rate was 400 mL/min, the dialysate flow rate was zero, and the ultrafiltration flow rate was 13 mL/min. UF samples were collected every 10 minutes from the UF line. Upon experiment completion at one-hour, residual filtrate was collected from the dialysate compartment. All samples were stored at −80°C until analysis. In total we conducted six dialysis simulations, each involving a concurrently tested dialyzer pair (Fig. 1B). UF samples underwent protein analysis (Bradford protein assay), albumin and creatinine measurement (Horiba® Pentra C400), liquid chromatography tandem mass spectrometry (LC-MS/MS)-based proteomics (Agilent® Q-TOF 6546), and immune-based protein assay (Luminex® Magpix). We analyzed the average of four sets of data obtained from the same dialyzer at the same time point. Results The loss of total protein and albumin was significantly greater with TH compared to the two HF dialyzers (Fig. 1C). Creatinine removal was similar across all dialyzers (Fig. 1D). The overall protein loss through F180NR and CorAL was comparable, while the protein loss through TH was up to 14-fold higher (811.7 ± 149.5 mg; Fig. 1E). Similar trends were observed for albumin. Albumin comprised 62% of the protein content in UF from TH, whereas it accounts for approximately 30% for HF dialyzers. We next employed label-free LC-MS and LC-MS/MS to identify and quantify proteins in 12 UF samples collected at the 20-minute point. Using Spectrum Mill software, we identified 244 proteins in the UF, 113 were accurately semi-quantified. The protein intensity ratios of TH-to- F180NR, TH-to-CorAL, and CorAL-to-F180NR for these 113 proteins are presented in Fig. 2A-C. Each gray dot on the plot corresponds to a specific protein. In line with the total protein and albumin quantification, MS proteomics analysis consistently revealed a greater loss of every identified protein by TH compared to both HF dialyzers. The median protein intensity ratio of TH-to-F180NR and TH-to-CorAL was 9.14 and 8.25, respectively. Importantly, TH not only removed middle-molecules such as β2M, but also showed a higher loss of salutary proteins. For instance, our data indicate a 9- to 16-fold difference in the level of albumin, haptoglobin, and vitamin D-binding protein between TH and HF dialyzers. Protein loss did not differ between F180NR and CorAL. To validate the accuracy of the MS quantification, we quantified seven proteins using immune-based assays in all UF samples at the 20-minute point. The result revealed a strong consistency between MS data and other methods (Fig. 2D). Conclusion We established an ex vivo system to concurrently test two dialyzers and implemented a robust proteomics workflow for a direct comparison of protein loss in the UF. Our data comprehensively and consistently indicate a greater extent of protein loss with MCO compared to HF, including the loss of salutary proteins. Further research is warranted to explore the clinical consequences of the enhanced loss of salutary proteins.