Thrombogenic characterization of alloyed and surface-modified magnesium bioresorbable metals for cardiovascular device applications

Abstract

Magnesium alloys show great promise for use in bioresorbable metal vascular stents due to their mechanical properties. However, considerable material engineering efforts are required to reduce the corrosion rate of magnesium and translate these stents into clinical use. Alloying elements and surface modifications are frequently used to reduce corrosion rate and retain mechanical strength for a longer time. This work sought to characterize the effects that these alloying approaches and surface modifications have on the acute thrombogenicity of magnesium scaffolds. Common magnesium alloys were assessed in conjunction with a biostable clinical control in both ex vivo whole blood and in vitro assays. Our results indicated that magnesium alloying did not affect the thrombogenicity of the material with equivalent platelet deposition and fibrin accumulation on all of the alloyed magnesium metals, as well as in the alloys’ proclivity to produce fibrin or activate factor XII (FXII). In contrast, surface modifications of magnesium, specifically fluorination and anodization, increased platelet deposition and fibrin accumulation onto the magnesium surface compared to the unmodified metal. No differences were found in fibrin or FXIIa generation between surface-modified and the unmodified magnesium material. In conclusion, this research demonstrated that alloying is a viable strategy to increase magnesium corrosion resistance without affecting its thrombogenicity, whereas surface modifications to magnesium may increase the material thrombogenicity.

Graphical Abstract

1. Introduction

Bioresorbable metal stents (BRMS) have the potential to overcome some limitations of permanent devices in vascular applications. A BRMS device, designed to dissolve after implantation in an ischemic artery, is consequently capable of reducing stent restenosis rates, stent thrombosis rates, and dual antiplatelet therapy (DAPT) treatment time.^1–3 Resorbable, biocompatible metals include iron, zinc, and magnesium that all have robust mechanical strength and ductility compared to polymers, enabling smaller struts, which reduce arterial injury and the risk of restenosis and thrombosis.⁴ Furthermore, the elution of metal degradation byproducts such as magnesium ions could be beneficial to patient vascular health.^5,6 In our recent work, we observed an anti-thrombotic effect of pure magnesium metal, supporting the material as a vascular stent biomaterial.⁷ Pure magnesium, however, corrodes too quickly for clinical use.⁸ To counter the rapid degradation, researchers alloy the bulk magnesium materials and modify the magnesium surface, such as with fluorination or anodization, to control the corrosion rate of magnesium devices.⁹ Without these modifications, the rapid corrosion of pure magnesium BRMS devices will lead to failure before the vasculature can heal.¹⁰

Particularly relevant to stent design is the level of biometal-mediated activation of the coagulation cascade. Current biostable stent scaffolds act as a permanent foreign surface in the body, leading to the initiation of the contact pathway of the coagulation cascade and the activation of platelets.¹¹ The contact pathway is initiated through the conversion of factor XII (FXII) into its activated form after contact with a charged surface, such as a stent. Activated FXII (FXIIa) represents an early step in the material-mediated coagulation cascade.¹² In prior work, we demonstrated that blocking FXII activation has a direct effect on reducing thrombus development in current biostable stents.¹³ FXII activation affects thrombus development through downstream conversion of fibrinogen into fibrin, the final product of the coagulation cascade, which is responsible for clot stability.¹⁴ Fibrin is a potent activator of platelets and is key to thrombogenesis.¹⁵ Activated platelets attach extended pseudopodia to fibrin, reinforcing the thrombus, while also releasing α-granules capable of activating inactive platelets and leukocytes.¹⁶ Using an acute thrombogenesis ex vivo model in tandem with investigating FXII activation and fibrin generation in vitro allows us to holistically examine the impact of alloying or surface-modifying magnesium on thrombogenesis.

The impact of magnesium corrosion on stent performance extends beyond losing mechanical integrity before the completion of adequate healing. Importantly, corrosion byproducts may also directly affect thrombus development through multiple means. Eluted magnesium ions act as cofactors for over 300 enzymes which can decrease inflammation, activate coagulation cascade proteins, and inhibit platelet activity.^17–20 Magnesium corrosion also releases hydrogen gas (H₂) to create an alkaline microenvironment, which has the potential to decrease coagulation protein adsorption and change the phenotype of platelets attached to the BRMS surface.^21–23 The alkaline microenvironment and release of bioactive magnesium ions are viable explanations for a magnesium device-induced reduction of platelet deposition.

This study aimed to determine whether alloying or surface-modifying magnesium alters the previously seen anti-thrombogenic effect of pure magnesium.⁷ We hypothesized that changing the magnesium surface, through either alloying or surface modification, would increase thrombogenesis compared to pure magnesium. We quantified the ability of alloyed and surface-modified magnesium materials to activate the coagulation cascade (FXIIa generation and fibrin formation) and induce thrombus (platelet and fibrin attachment) formation. We also investigated the effects of magnesium’s corrosion byproducts, ion release or induced pH change, as the probable causes of the reduction in thrombogenesis for magnesium materials. Our results indicated that alloying magnesium had no effect on thrombogenicity whereas surface-modification increased thrombogenesis in whole blood. Eluted byproducts (pH and magnesium ions) affected thrombogenicity differently, with pH abolishing the contact pathway depending on alkalinity, and magnesium ions having little effect on the contact pathway.

This work is vital to furthering device design – discovering the key features of the metal surface that define thrombogenesis can offer researchers specific goals for material development. Furthermore, discovering the mechanism of reduced thrombogenicity in some metals, like magnesium, could offer a pathway to clinical use of a bioresorbable, non-thrombogenic metal stent.

2. Materials and Methods

2.1. Materials

Alloyed magnesium wires (ZX10, AZ31, WE43) and surface-modified wires (WE22, WE22-A, WE22-HF) were supplied by Fort Wayne Metals (Fort Wayne, IN); biostable clinical controls (CoCr) and pure magnesium (99.9%) were supplied by Goodfellow Corp (Pittsburgh, PA). All wires were cut into straight 6 mm segments for in vitro testing or cut to 12.7 cm lengths and coiled around a 4mm Teflon rod for ex vivo experiments. Materials were stored in 200 proof ethanol and sonicated for 10 minutes, then rinsed in dPBS prior to testing. The magnesium alloys are described: ZX10 (Mg-1Zn-0.3Ca-0.15Mn), AZ31 (Mg-3Al-1Zn-0.5Mn), WE43 (Mg-4Y-3Nd-0.3Zr), and WE22 (Mg-2Y-1.5ND-0.5Zn). The surface modifications of WE22 were performed by Fort Wayne Metals and included anodization (WE22-A) or soaking in hydrofluoric acid (WE22-HF). The clinical control was a CoCr alloy formulation of Co40-Cr20-Fe15-Ni15-Mo7-Mn2-C-Be. All wires were 0.25 mm diameter. Pure magnesium was as drawn, CoCr was hard worked, and all magnesium alloys were annealed.

2.2. Whole blood thrombogenesis

Acute thrombus formation studies were performed in non-human primate (NHP) studies at the Oregon National Primate Research Center. All studies were approved by the on-site Institutional Animal Care and Use Committee. These experiments followed guidelines set by the National Research Council and the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources. This model has been described in our prior work, with a summary and additional details below.^7,13 Importantly, our work utilizes whole blood flowing ex vivo without anticoagulants or antiplatelet therapies, allowing for assessment of thrombogenesis without conflicting therapies.

2.2.1. Radio-labelled platelets and fibrin for thrombus quantification

Materials for the ex vivo shunt loop were 12.7 cm length wires coiled around a Teflon rod with an outer diameter of 4mm. After coiling, the final length was 2 cm long. The coils had fluorinated ethylene propylene (FEP) heat-shrink tubing (Zeus Inc., Orangeburg, SC) around the outside of the coil and shrunk to 4 mm inner diameter (ID) through applied heat. These segments of FEP-tubing with coils were added into the larger shunt loop utilizing Silastic tubing (4 mm ID diameter, Dow Chemical).

Arteriovenous shunts were implanted in one healthy male Papio anubis baboon (12 kg) based on our established model.^7,24–27 The magnesium alloy testing was performed in the right leg’s arteriovenous shunt. One month after completion of those studies, the left leg shunt was placed and the surface-modified WE22 metals were tested. Platelets and fibrinogen were radio-labelled with isotopes 111-In and 125-I, respectively, for quantification. These autologous platelets were imaged each minute during the shunt loop using a Brivo NM615 (General Electric, Boston, MA). Blood flowed through the tubing across the different coils without anticoagulants for one hour. The flow rate was controlled to 100 mL/min using a clamp distal to the coil, resulting in a shear rate of 265 s⁻¹ (Fig. 1A). After 1 hour, the coil was rinsed with saline and an image was taken to macroscopically observe the thrombus (Fig. 2A). After the decay of 111-In, coil segments were examined for fibrin quantification using a 1480 Wizard gamma counter (PerkinElmer, Waltham, MA).

Figure 1:

Whole blood testing in ex vivo shunt and SEM images of the as-received wires. A) Macroscopic images of the NHP shunt loop during data collection, where the red arrow denotes the direction of blood flow and the blue arrow indicates the site of downstream whole blood sampling (DSS). Finally, the green arrow indicates the site of samples used in SEM imaging. B) SEM images of the as-received wires prior to whole blood testing in the shunt. Scale bar = 50 μm.

Figure 2:

Magnesium alloy thrombogenicity was tested in whole blood ex vivo (2A-2F) and in plasma in vitro (2G-2I). A) After 1 hour of exposure to non-anticoagulated whole blood, minimal thrombus formed on all magnesium alloys with an accumulation of erythrocytes, leukocytes, and plasma protein adsorbed to the surface of every material. ZX10, WE43, and AZ31 had similar levels of visible blood component deposition, while CoCr was entirely coated with thrombus elements. Scale bar = 50 μm. B) After 1 hour of exposure to non-anticoagulated whole blood, magnesium alloys ZX10, WE43, and AZ31 had significantly lower platelet deposition onto the surfaces compared to the CoCr (p < 0.001, n=5). C) After 1 hour of exposure to non-anticoagulated whole blood, ZX10 and WE43 had significantly lower fibrin accumulation on their surfaces compared to CoCr (p < 0.01). AZ31 trended toward have less fibrin accumulation than CoCr (n=5). D) From the blood samples drawn in the boundary layer downstream of the materials, magnesium materials did not generate statistical differences between alloys in MPO concentration after 1 hour of exposure to non-anticoagulated whole blood. MPO did increase significantly with time (0 min. vs. 30 min., p < 0.01; 0 min. vs. 60 min., p < 0.01) but with no significant difference between alloys (n=5). E) TAT generation, from boundary layer blood, was not statistically different between materials within 1 hour. TAT concentration did significantly increase between time points during the 1 hour (0 min. vs. 30 min., p < 0.0001; 0 min. vs. 60 min., p < 0.0001, n=3–5). F) PF4 concentration, from boundary layer blood, had no statistically significant difference between different materials. PF4 concentration increased significantly after exposure to all materials between time points (0 min. vs. 30 min., p < 0.0001; 0 min. vs. 60 min., p < 0.0001, n=3–5). G) After 1 hour, all magnesium alloys in human plasma generated significantly lower FXIIa compared to the CoCr clinical control (p < 0.0001). WE43 had a significantly lower rate of FXIIa cleavage than pure magnesium (p < 0.01, n=9). H) The time to initiate fibrin clotting by every alloy in plasma was comparable to plasma alone. Pure magnesium (p < 0.05) and AZ31 (p < 0.01) caused significantly longer clotting times than the CoCr clinical control (n=9). I) There was no difference in rate of fibrin generation caused by the materials (n=9). For the fibrin deposition, platelet accumulation, and in vitro tests, data were analyzed with a one-way ANOVA and Tukey’s post hoc. TAT and PF4 concentrations were log transformed to achieve normality assumptions. For TAT, PF4, and MPO assays, data were analyzed with a two-way ANOVA or a mixed effects analysis (2E & 2F) and Tukey’s multiple comparisons test. * = p < 0.05, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001.

2.2.2. SEM of materials after whole blood exposure

Straight 2 cm wires were placed into a 4 cm inner diameter Silastic tubing across the flow field. FEP tubing was placed on either end to secure the wire. Samples were placed into the shunt loop approximately 0.8 m downstream of the coil (Fig. 1A). After the 1 hour loop and saline rinse, these straight wires were rinsed in PBS and fixed in 3.7% paraformaldehyde overnight at 4°C. These SEM samples were dehydrated in escalating concentrations of ethanol and stored in 200-proof ethanol. The final dehydration was performed overnight with hexamethyldisilazane (HDMS, Sigma, St. Louis, MO). Dehydrated samples were mounted onto an SEM stud using carbon tape and sputter coated with gold (ACE600 Leica). Base-line wires without blood contact were affixed to a SEM stud via carbon tape. These samples were imaged without sputter coating. (Fig. 1B). All samples were imaged with a Volumescope 2 SEM (Thermo Scientific, Waltham, MA) with high vacuum.

2.2.3. Blood testing downstream of materials

Downstream of the tested coils, boundary layer blood was drawn into citrate (10% by volume), based on previous methods.7 This boundary blood layer draw enabled interrogation of the surface-blood interaction.²⁸ Prior to the start of the ex vivo study (labelled “0 min”), blood was bulk drawn into 10% by volume citrate. Boundary blood draws at 1.5 mL/h were taken from 0 to 30 min (labelled “30min”) and from 30 to 60 min (labelled “60min”). After collection, all blood was centrifuged for 3 minutes at 10,000 RPM to isolate platelet poor plasma. Plasma samples were stored at −80°C until used in the biochemical assays.

Plasma thrombin anti-thrombin (TAT) concentration was measured using the Siemens Healthineers (Erlangen, Germany) TAT ELISA kit. Platelet-factor 4 (PF4) in downstream plasma was measured using a PF4 ELISA kit from R&D Systems (Minneapolis, Minnesota). Plasma myeloperoxidase (MPO) activity was measured using a fluorescent detection kit from Cell Technology (Hayward, California). All kits were run according to the manufacturer’s instructions. Plasma for TAT ELISA was diluted 1:4, 1:16, and 1:256 (plasma:diH₂O) for the 0 min, 30 min, and 60 min time points, respectively. Plasma for was diluted 1:1000 for PF4, and 1:1 for MPO with dPBS. In the colorimetric TAT and PF4 assay, data points were excluded if plasma was visibly pink from thrombus or hemolysis.

2.3. Activation of the contact pathway in coagulation

The ability of the bioresorbable metals with and without coatings to activate the coagulation cascade was evaluated with in vitro testing of wires and plasma using methods similar to prior work,⁷ with a brief description and modifications noted below.

2.3.1. FXII activation

Straight wires were placed in a non-tissue coated 96-well plate and rinsed twice with Dulbecco’s phosphate buffered saline (dPBS). Human plasma [Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis (SSC/ISTH) Coagulation Standard] was reconstituted at room temperature with 1 mL DI water and diluted 1:1 with 37°C dPBS. Plasma was kept on a 37°C heat block during the experiment. Wire samples were submerged in plasma and a chromogenic substrate for activated FXII, Chromogenix S-2302 (2:1 plasma to reagent). After the addition of the chromogenic substrate, the plate was immediately placed in a plate reader (infinite M200 spectrophotometer) for a 1 hour dynamic assay. The plate was maintained at 37°C temperature for the assay and absorbance measurements of 405 nm wavelength were taken every minute. The rate of color change was used to directly assess the quantity of activated FXII in solution.

2.3.2. Initiation and rate of fibrin generation

Straight 6 mm wires were placed in a non-tissue coated 96-well plate and rinsed twice with dPBS. SSC/ISTH plasma was reconstituted at room temperature with 1 mL DI water and diluted 1:1 with 37°C dPBS. Plasma was kept on a 37°C heat block during experiment. Plasma was added to the wells with materials, and then 25 mM CaCl₂ was added for a final concentration of 8.3 mM CaCl₂. The plate was put into the plate reader (infinite M200 spectrophotometer) immediately after the addition of CaCl₂ for a dynamic assay of 1 hour. The plate reader was programmed to maintain 37°C temperature during the assay and to take absorbance readings at 405 nm wavelength every minute. The time for solution turbidity change (10% over baseline) was recorded as the fibrin clotting time. The rate of changing turbidity was recorded as the rate of fibrin generation.

2.4. Role of magnesium corrosion byproducts in activation of the contact pathway in coagulation

2.4.1. Corrosion byproduct quantification

To quantify the degradation of the alloyed and surface-modified magnesium materials with the eluted magnesium ion content, inductively coupled plasma mass spectrometry (ICP-MS) was performed. 6 mm wires of magnesium (99.9%), magnesium alloys (ZX10, WE43, AZ31, WE22) and surface-modified WE22 wires (WE22-A, WE22-HF) were incubated in 100 uL of SSC/ISTH plasma for 1 hour in a 37°C incubator. Plasma was then collected and analyzed with ICP-MS to determine magnesium ion content eluted from the wire. Negative controls included an empty tube (no solution) and plasma alone to determine contaminating and baseline quantities of metals, respectively, which were then used to determine the increase of ion content from baseline. After exposure to the materials for 1 hour, the pH of the plasma was recorded with a pH probe (ThermoScientific Orion™ Versa Star Pro^™).

2.4.2. Impact of corrosion byproducts on activation of the contact pathway in coagulation

Magnesium corrosion causes both an increase in [Mg²⁺⁺] and a rise in pH of plasma both of which may impact coagulation. To examine the effect of increased [Mg²⁺⁺], a MgSO₄:PBS solution was added to SSC/ISTH plasma to raise plasma [Mg²⁺⁺] to 2 mM. This value is based on the ICP-MS experimental results. The impact of the magnesium ion-modified plasma on coagulation was compared to unmodified plasma and the same magnesium wire using the methods of the activation of contact pathway in coagulation experiments (section 2.3). Furthermore, the clinical CoCr control was also incubated with 2mM magnesium ion-modified plasma to quantify possible inhibition of thrombogenesis by magnesium ions independent of the magnesium material surface. To quantify the impacts of pH on coagulation, SSC/ISTH plasma was aliquoted, and the initial pH was measured using a pH probe (ThermoScientific Orion™ Versa Star Pro™). The plasma pH was raised to either pH 9 or pH 10 using 1 M NaOH. These plasma conditions were compared to baseline plasma at its normal pH (pH 7.9, S1A) and a magnesium wire using the methods of the activation of contact pathway in coagulation experiments (section 2.3). To ensure the fidelity of these coagulation assays, we tested increasing alkaline plasma without materials to ensure accuracy (Fig. S1B & S1C). For testing the FXII assay (Fig. S1B), we used 200nM FXIIa in all conditions to ensure changes in color were not being driven by altering the pH.

2.4. Statistics

All data are presented as mean ± standard deviation. In vitro data were tested for analysis of variance (ANOVA) assumptions, using Shapiro-Wilks and Brown-Forsythe tests with quartile-quartile plots. In vitro tests were done with three replicates, in triplicate (n=9). Data sets that passed ANOVA assumptions were analyzed with a 1-way ANOVA and a Tukey’s post hoc was performed if the main effects test showed significance (p < 0.05). A Greenhouse-Geisser correction was used in all ANOVAs. Whole blood testing of each material was performed in the shunt loop five times. Downstream sampled plasma was then used for TAT, PF4, and MPO biochemical assays (n=5). For MPO data, a 2-way ANOVA with factors of time and metal type was utilized. TAT and PF4 ELISAs had missing data points due to clotting in the sample and were analyzed with a mixed-effects model. Additionally, TAT and PF4 failed normality assumption tests and were log transformed to reach required assumptions for analysis. Data are presented in their transformed version. Data sets were analyzed with GraphPad Prism version 11 in conjunction with R for assumption testing. No data were excluded from in vitro fibrin clotting time and FXIIa generation assays. In the PF4 and TAT experiments, data were excluded when the concentration either exceeded or fell below the limits of the calibration curve. Additionally, if clotting or hemolysis was present in the plasma sample (visibly changing the serum color), that sample was excluded from the ELISA.

3. Results

3.1. Magnesium alloys

3.1.1. Thrombogenesis in whole blood

Imaging and quantification of thrombus components after 1 hour of exposure to flowing whole blood in an ex vivo shunt loop, showed differences between the magnesium alloys compared to the CoCr clinical control (Fig. 2A–2C). Representative macroscopic images of the coils indicated a larger thrombus on the CoCr clinical control compared to the three alloys. Further physical examination was performed on the wires using SEM (Fig. 2A). The presence of erythrocytes, leukocytes, platelets, and fibrin was confirmed on all wires imaged. Radio-labelled platelets were quantified via a gamma camera. ZX10, AZ31, and WE43 all had significantly less (p < 0.001) platelet accumulation on their surface compared to the biostable CoCr clinical control (Fig. 2B). There was no statistically significant difference of platelet accumulation between the magnesium alloys and the negative control of no metal coil. Additionally, radio-labelled fibrin was quantified as a single endpoint. Both ZX10 and WE43 had significantly lower fibrin accumulation on the surface than the CoCr clinical control (p < 0.01) (Fig. 2C).

Using plasma isolated from downstream boundary layer blood samples, immune activity was assessed by quantifying MPO concentration, while TAT and PF4 concentrations were quantified to interrogate material-induced activation of coagulation and platelets, respectively. For the magnesium alloys, the materials did not significantly alter the MPO content in the plasma (Fig. 2D); however, the magnesium alloys did cause a significant increase of MPO concentration over time (p < 0.001) (Fig. 2D). Similarly, the magnesium alloys did not significantly alter the plasma concentration of TAT and PF4 while all materials caused a significant increase in TAT and PF4 concentration over time (p < 0.001) (Fig. 2E–2F).

3.1.2. Activation of the contact pathway of coagulation

To determine material effects on the coagulation cascade, the magnesium alloys’ impact on the activation of FXII and the initiation of fibrin were tested in vitro. The magnesium alloys caused significant differences in FXII activation and fibrin clotting time (Fig. 2G–2I). The WE43 alloy (as well as plasma only control) induced significantly lower FXIIa substrate cleavage rate compared to pure magnesium (p < 0.05) (Fig. 2G). All materials had significantly lower levels of FXIIa than the CoCr clinical control (p < 0.0001). The time to initiate fibrin clotting was significantly delayed for the pure magnesium (p < 0.05) and AZ31 material (p < 0.001) compared to CoCr (Fig 2H). No magnesium alloy caused a significantly different fibrin clotting time compared to plasma alone. All materials, as well as the clinical and negative control, did not significantly alter the rate of fibrin generation (Fig. 2I).

3.2. Magnesium surface-modifications: anodization or fluorination

3.2.1. Thrombogenesis in whole blood

Imaging and quantification of thrombus components after 1 hour of the shunt loop showed differences between surface-modified WE22 wires (Fig. 3A–3C). Representative images of the coils after the final rinse indicated more thrombus deposition onto the surface-modified WE22 groups than bare WE22 (Fig. 3A). SEM showed less visible deposition of fibrin, erythrocytes, leukocytes, and platelets on bare WE22 when compared to the surface-modified alloys. Bare WE22 wires had significantly lower platelet deposition than WE22-A and WE22-HF (p < 0.001) during exposure to whole blood (Fig. 3B). Additionally, WE22 had significantly lower fibrin accumulation than WE22-HF (p < 0.05), while trending lower compared to WE22-A (Fig. 3C). Importantly, WE22-A differed from other materials by having loosely adhered clots – three of the five WE22-A samples had clots break off the wire coil into the bloodstream, which is represented in the data by a reduction in platelet deposition near the 50 minute mark (Fig. 3B).

Figure 3:

WE22 magnesium alloy, with and without surface modifications, was tested ex vivo (3A-3F) and in vitro (3G-3I). A) After 1 hour of exposure to non-anticoagulated whole blood, minimal thrombus formed on unmodified WE22, with considerable thrombus on WE22 after anodization (WE22-A) or fluorination (WE22-HF). SEM images showed accumulation of erythrocytes, leukocytes, and plasma protein adsorbed to the surface. WE22 had visibly less deposition of thrombus elements compared to WE22-A, WE22-HF. Scale bar = 50 μm. B) After 1 hour of exposure to non-anticoagulated whole blood, WE22 had significantly lower platelet deposition on its surface compared to both of the surface-modified metals, WE22-A and WE22-HF (p < 0.001, n=5). Three of five of the thrombi on WE22-A samples detached from the wires near the end of the study, which caused an average decrease in platelet concentration around the 50-minute mark. Each individual run can be seen in Supplemental Figure 2. C) After 1 hour of exposure to non-anticoagulated whole blood, WE22 had significantly lower fibrin deposition than WE22-HF (p < 0.05), but there was no significant difference found in fibrin deposition compared to WE22-A (n=5). D) From the blood samples drawn in the boundary layer downstream of the materials, surface modifications did not significantly alter MPO concentration between materials, but MPO did increase significantly with time (0 min. vs. 30 min., p < 0.001; 0 min. vs. 60 min., p < 0.001) but with no significant difference between alloys (n=5). E) TAT generation, from boundary layer blood, was not statistically different between unmodified and surface-modified materials within 1 hour. TAT concentration did significantly increase between time points during the 1 hour (0 min. vs. 30 min., p < 0.0001; 0 min. vs. 60 min., p < 0.0001, n=3–5). F) PF4 concentration, from boundary layer blood, had no statistically significant difference between different materials. PF4 concentration increased significantly after exposure to all materials between time points (0 min. vs. 30 min., p < 0.001; 0 min. vs. 60 min., p < 0.001, n=3–5). G) After 1 hour in human plasma, the CoCr clinical control caused a significantly higher rate of FXIIa substrate cleavage compared to any other samples (p < 0.0001, n=9). H) The time to initiate fibrin clotting by unmodified and surface-modified WE22 in plasma was comparable to plasma alone and significantly longer than clinical control CoCr (p < 0.05, n=9). I) There were no significant differences in fibrin generation rate based on material surface treatment (n=9). For the fibrin deposition, platelet accumulation, FXIIa generation, and fibrin clotting times, stats were analyzed with a one-way ANOVA and Tukey’s post-hoc test. TAT and PF4 concentrations were log transformed to reach normality assumptions. For TAT, PF4, and MPO assays, data were analyzed with a two-way ANOVA or a mixed effects analysis (3E) and Tukey’s multiple comparisons test. * = p < 0.05, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001.

Each individual run can be seen in Supplementary Data with the corresponding fibrin weight to illustrate how the clot embolization decreased endpoint fibrin data (Fig. S2).

The plasma isolated from downstream boundary layer blood samples was used to quantify material-induced immune activity (MPO) and activation of coagulation and platelets (PF4 and TAT). Surface-modified WE22 materials caused a significant increase in MPO concentration (p < 0.0001), PF4 concentration (p < 0.001), and TAT concentration (p < 0.0001) in all materials over time (Fig. 3D–3F). Surface modification of WE22 had no effect on the downstream plasma samples in regard to MPO concentration, TAT concentration, and PF4 concentration.

3.2.2. Activation of the contact pathway of coagulation

Surface-modified WE22 and bare WE22 caused no differences in FXII activation or fibrin clotting time (Fig. 3G–3I) in human plasma in vitro testing. All materials had significantly lower FXIIa substrate cleavage rates compared to CoCr clinical control (p < 0.0001) (Fig. 3G). All WE22 materials produced significantly longer clot initiation times compared to CoCr (p < 0.01) (Fig. 3H). Finally, no surface-modified WE22 wire caused a significant difference in fibrin generation rate compared to any other material or controls (Fig. 3I).

3.3. Impact of magnesium corrosion byproducts on the contact pathway of coagulation

Functional corrosion testing was performed to determine in situ byproduct generation and the byproducts impact on material induced activation of the coagulation cascade. The corrosion was investigated through ICP-MS, quantifying the concentration of released magnesium ions into plasma (Table 1). Similarly, pH of wires incubated with magnesium wires was recorded (Supplementary Data, Fig. S1). This change in pH was used to inform static condition experiments. Additionally, the FXII activation and fibrin clotting time assays were repeated in alkaline plasma with no materials to ensure their fidelity (Supplemental Data, Fig. S1B & Fig. S1C). The FXIIa experiment utilized a set concentration of 200nM FXIIa with varying pH to ensure color change was not being affected by pH alone. Importantly, there was a small but statistically significant increase in FXIIa from the alkaline plasma pH 9.0 compared to pH 7.4 plasma. This increase was too small to have an effect on the low amounts of FXII activation by magnesium materials in vitro.

Table 1:

Pure magnesium, magnesium alloyed, and surface modified magnesium wires were incubated in human plasma for 1 hour at 37°C. ICP-MS was used to quantify elemental content. Magnesium samples caused variable ion release into solution while plasma alone maintained a similar concentration across all three trials. WE22-HF and plasma alone had significantly less magnesium ions in solution (* = p < 0.05) than the other materials. Data were analyzed with a one-way ANOVA (n=3), and data are shown as mean +/− the standard deviation. Additional elemental content analyses are provided in Supplemental Table 1.

Material	Composition	Mg Ion Concentration (ppm)
Pure Magnesium	99.9% Mg	107.2 ± 42.39
ZX10	Mg-1Zn-0.3Ca-0.15Mn	86.3 ± 23.92
WE43	Mg-4Y-3Nd-0.4Zr	109.3 ± 41.41
AZ31	Mg-3Al-1Zn-0.5Mn	73.0 ± 19.45
WE22	Mg-2Y-1.5Nd-0.5Zn	125 ± 7.90
WE22-HF	Mg-2Y-1.5Nd-0.5Zn	31 ± 4.21*
WE22-A	Mg-2Y-1.5Nd-0.5Zn	89.8 ± 2.44
Plasma	No material	8.2 ± 0.47*

All materials caused an increase in magnesium ion concentration compared to the no material plasma control (Table 1). During the 1 hour incubation period, the magnesium ions eluted into human plasma by the magnesium alloys (Mg, ZX10, WE43, and AZ31) were not significantly different from each other – ranging from a 10.5X (ZX10) to a 15.2X (WE22) increase in magnesium ion concentration compared to baseline plasma. Surface modification of WE22 had varied effects on ion release, with WE22-HF having significantly lower (p < 0.05, 4X increase from plasma alone) magnesium ion release than all other materials. Alternatively, WE22-A caused a 10.9X increase in magnesium ion concentration compared to plasma alone. Ion concentration of elements other than magnesium can be found in Supplementary Data (Table S1).

Human plasma treated with magnesium ions, the magnesium wire, and untreated plasma were not significantly different in FXII activation (Fig. 4A). While the CoCr clinical control increased FXII activation compared to the other samples, the CoCr in magnesium ion-treated human plasma was not significantly different than the CoCr in plasma alone. For the fibrin clotting time assay, there was no significant difference between baseline plasma, plasma with a magnesium wire, or plasma with an increased magnesium ion concentration (Fig. 4B). The CoCr clinical control in magnesium ion treated plasma had a longer average clotting time than the CoCr clinical control in baseline plasma; however, this was not statistically significant (p = 0.058). There were no differences in the fibrin generation rate for all groups in the magnesium ion plasma study (Fig. 4C). As expected, all material conditions and the magnesium ion treated plasma had significantly lower FXII activation and a longer fibrin clotting time than the CoCr clinical control.

Figure 4:

In vitro assays tested magnesium elution products to determine anti-thrombogenicity effect. A) There was no significant difference between magnesium ion treated plasma and native plasma. Similarly, there was no difference in clotting time between CoCr in native plasma or magnesium ion treated plasma. Plasma, magnesium material, and magnesium ion treated plasma had significantly less FXII activation than CoCr in either native plasma or magnesium ion treated plasma (p < 0.0001, n=9). B) There was no significant difference in fibrin clotting time between plasma alone and magnesium ion treated plasma. Similarly, there was no significant difference between CoCr clotting time in native plasma compared to CoCr in magnesium ion treated plasma, though the latter had an increased average clotting time approaching significance (p = 0.058). Plasma alone, the magnesium wire, and the magnesium ion treated plasma all had significantly longer clotting times than CoCr. (p < 0.001, n=9) C) Fibrin generation rate had no significant difference between plasma with magnesium material or Mg²⁺⁺ solution (n=9). D) Magnesium material had a similar reduction in FXIIa cleavage rate compared to pH 9 plasma, while pH 10 plasma had no detectable FXIIa. All conditions had a significantly lower FXIIa generation rate than CoCr (p < 0.0001, n=9). E) All tested materials had significantly shorter clotting times than the pH-altered plasma since the pH-raised plasma did not clot during the entire 1 hour assay (p < 0.0001). The clinical control, CoCr had significantly shorter clotting times compared to plasma alone and plasma with the magnesium material (p < 0.001, n=9). F) The lack of fibrin clotting led to a significant increase of fibrin generation rate in material conditions compared to alkaline plasma (p < 0.0001, n=9). All analysis was performed via a one-way ANOVA with a Tukey’s post-hoc test. * = p < 0.05, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001.

There were no significant differences in FXII activation between plasma with elevated pH (pH 9 and pH 10), plasma containing magnesium wire, and baseline plasma (Fig. 4D). Baseline plasma and material groups (including CoCr clinical control) had significantly (p < 0.0001) shorter fibrin clotting initiation time compared to the alkaline plasma conditions; there was never a successful clot in either the pH 9 or pH 10 condition (Fig. 4E). Due to this failure to clot, the alkaline plasmas caused statistically significant differences in the fibrin generation rate compared to all materials and the negative control (p < 0.0001). (Fig. 4F). It should be noted that the significance comes from a value assigned to the “no clot” conditions: 3600 seconds for the clotting time assay, which represents the 1 hour the assay runs, (Fig. 4E) and a slope of 0 for fibrin generation rate (Fig. 4F).

4. Discussion

This work sought to characterize the effects of alloying elements and surface modifications on magnesium’s thrombogenicity. Magnesium, magnesium alloys, and surface-modified magnesium materials were tested in whole flowing blood as well as human plasma. Additionally, the impacts of byproducts of magnesium degradation were tested for mechanistic insight. This tripartite approach ties the whole blood response to fundamental analyses that separately identify the effects of corrosion byproducts and magnesium surface on thrombogenesis. Overall, alloying elements had little effect on magnesium thrombogenicity, as all alloys had similarly low levels of platelet and fibrin accumulation, FXII activation, and fibrin clotting time. This was not the case for surface-modified magnesium materials. In the whole blood thrombogenesis model, surface modification increased thrombogenicity, whereas the contact pathway of coagulation benchtop testing in human plasma showed no differences between materials. The downstream blood samples from the boundary layer of whole, flowing blood in contact with the materials were analyzed for myeloperoxidase (MPO), thrombin-anti-thrombin (TAT) complex and platelet factor 4 concentration (PF4) and there were no significant differences between any material composition or surface treatment. To better understand the changes in the coagulation cascade, the common elution byproducts of magnesium corrosion were used to modify human plasma. The effect of byproducts tested (pH and Mg²⁺⁺) varied; plasma with increased magnesium ion concentration had the same fibrin clotting time and FXIIa generation as baseline plasma, whereas alkaline plasma abolished fibrin clotting time and altered FXII activation. This assessment and the increased thrombogenicity of surface-modified materials in whole blood testing indicate that the corrosion process affects platelet deposition and thrombus development. This work sought to understand the relationship between the early events of thrombogenesis and magnesium surfaces, modified either through different alloying elements or surface modifications. Furthermore, the experiments presented in this work bring us closer to identifying the underlying cause of magnesium’s anti-thrombogenic effects, recorded in prior research.⁷

Magnesium is commonly alloyed with different metals (e.g., aluminum and zinc) to improve mechanical properties and corrosion resistance.²⁹ The effect of these alloying elements on magnesium’s thrombogenicity is poorly understood. Common methods of evaluating alloy thrombogenicity include quantifying platelet activation and deposition, as platelets are key mediators of thrombogenesis.^30–32 Gu et al. found that binary alloys of magnesium with various elements decreased hemolysis and reduced static platelet deposition compared to pure magnesium; however, there was no statistical significance found in the study.³³ Furthermore, Yahata and Mochizuki found that alloying elements (aluminum and zinc) caused decreased platelet activation compared to pure magnesium.³⁴ Our hypothesis was that alloying magnesium would alter acute thrombogenesis in the whole blood ex vivo model; however, this was not the case. Testing with this clinically relevant flowing whole blood model demonstrated no effect on thrombogenesis from alloying magnesium in the context of platelet deposition. Beyond platelets, our work also displayed no differences between alloys in the context of the contact activation pathway of the coagulation cascade. This pathway is activated when FXII encounters a foreign charged surface and becomes FXIIa, which leads downstream to thrombin generation and the resulting conversion of fibrinogen to fibrin.³⁵ The reduction of FXII activation and fibrin clotting time compared to the CoCr clinical control in our stagnant plasma experiments, in tandem with reduced fibrin deposition in the whole flowing blood experiment, indicated that magnesium and magnesium alloys could reduce thrombogenesis through reduced activation of the contact pathway of the coagulation cascade. The absence of differences seen between materials in our experiments suggest that the bulk material is more important than the alloying elements in the context of acute thrombogenesis; however, there could be differences that were not seen in these studies. Namely, understanding the nature of the proteins within the thrombus, by using high-fidelity methods such as proteomics, could point to constitutive differences in alloy thrombogenesis. Additionally, understanding the availability of various elements on the magnesium alloy surfaces could reveal element specific effects on thrombus development. Future work should tie spatial analysis of the material surface with in-depth proteomic quantitation of thrombus elements to elucidate possible differences in thrombogenesis between these materials.

Surface-modification is an additional strategy used to reduce magnesium and magnesium alloy corrosion in vivo.^36,37 Two common surface-modification strategies were tested in this study: hydrofluoric acid modification and anodized oxidization. Hydrofluoric acid modification, in which the metal is coated with a stable layer of magnesium fluoride (MgF₂), protects from aqueous corrosion. Previous work has found that fluorination of magnesium alloys increased cytocompatibility; however, the effects of thrombogenesis are not known.^38,39 Anodized oxidation, or anodization, creates a stable metal oxide surface on magnesium materials to prevent corrosion. There is some evidence that anodization does not alter thrombogenicity: researchers found no changes in hemolysis or static thrombin generation between anodized and commercially pure magnesium in static human blood hemocompatibility tests.^40,41 We hypothesized that the modified surface would alter the thrombogenic response. Our work found significant and possibly severe changes in acute thrombogenesis when the WE22 material was modified. Both surface-modifications increased fibrin deposition and platelet accumulation in the whole blood ex vivo model. The differences in the results of our work compared to prior experiments testing the effect of surface-modification on thrombogenesis can be attributed to our use of non-anti-coagulated whole blood, rather than static testing of anti-coagulated human blood. Beyond the increased thrombogenesis, WE22-A induced the formation of loose clots that embolized into the system. This is a novel effect of the WE22-A material that was not been seen in the magnesium alloys or other surface modification of WE22, indicating a unique relationship between the anodized surface and thrombus elements.^13,28,41 The embolization of thrombus elements seen in WE22-A could be a clinically relevant event; however, the prevalence of dual anti-platelet therapy in clinical applications would greatly alter thrombogenesis.⁴² While our work grants insight into the macroscopic effect of surface-modification on thrombogenesis, it does not explain the mechanism behind the embolization. Future research using whole blood thrombogenesis testing should look at alterations in anodization parameters, such as thickness or voltage, to examine if the blood clot embolization effect can be reduced. Both surface modifications increased thrombogenesis in the whole blood ex vivo experiments; however, there was no effect of these surface modifications on the contact pathway coagulation cascade. This provides evidence that the increase in acute thrombogenesis is not mediated by FXII; however, there are key differences in whole blood that may also explain the opposing results. Namely, the static plasma experiments on the activation of the contact pathway of the coagulation cascade lack platelets, replenishment of fibrinogen, and the hemodynamic flow of blood – all of which can dramatically alter thrombogenesis and thrombus progression.^44,45

Magnesium’s high corrosion rate and bioactive corrosion byproducts could explain the low level of thrombogenesis seen in these experiments. Early studies described increasing magnesium ion concentration reducing clotting time in vitro and in vivo.⁴⁶ Magnesium ions potentiate Factor IX and Factor VIII, additional components of the coagulation cascade, but there is no consensus on its effect on FXII.^20,47 We hypothesized that increased [Mg²⁺⁺] would decrease coagulation cascade activation; however, this was not the case. First, our ICP-MS data showed no statistically significant differences in magnesium ion elution between magnesium alloys and pure magnesium in human plasma. Only WE22-HF displayed reduced corrosion compared to other materials and this difference had no effect on its ability to activate FXII or form a clot. Our experiments indicated that magnesium ions had no effect on reducing FXII activation or on increasing the fibrin clotting time when compared to the baseline plasma or the magnesium wire. In fact, the only effect of magnesium ions on the coagulation cascade was non-statistically significant increase in fibrin clotting time for the CoCr clinical control when incubated with magnesium ion treated plasma compared to baseline plasma. This indicates little effect of the magnesium ion released during biomaterial degradation on the contact pathway of the coagulation cascade, as neither the initiating protein, FXII, nor the terminating protein, fibrin, was significantly altered by its presence in plasma. Another byproduct of magnesium that may affect thrombogenesis is eluted hydroxides (OH−) and hydrogen gas that create an alkaline environment.²¹ pH has been known to alter thrombogenicity, either through altering thrombin generation or the platelet response.^22,23 We hypothesized that the pH increase would alter the contact pathway of coagulation. Our studies found that pH strongly altered FXII activation and fibrin clotting time – at pH 10, FXII activation was almost abolished, with most trials (5 of the 9) displaying no FXII activation. It is possible that the alkaline pH prevents FXII activation through transient deformation or even complete denaturation; however, further work would be required to confirm the specific effect on FXII. Similarly, fibrin was not generated at bulk pH values greater than 8.5. While these experiments indicated an effect of alkaline pH on the contact pathway, it’s difficult to explain these changes as the cause of reduced thrombogenesis in whole blood. These static experiments have a higher pH change than the in vivo reality of magnesium stents, due to a lack of blood flow and limited plasma volume. Additionally, while pH was elevated in the pure magnesium wire conditions, FXII activation and fibrin clotting did occur, indicating that the surface of the material can still cause thrombosis in spite of these pH changes. Lastly, it is unknown whether this bulk change in pH accurately captures the alkaline microenvironment that would exist for magnesium stents in vivo. Microprobe based pH studies of magnesium alloys in static simulated body fluid (Hank’s Salt Solution with additional calcium) found different changes in alkalinity: one study showed a minor increase at the surface from 7.4 to <8.0, whereas another reported a more substantial increase from 7.4 to ~8.5. ^48,49 Importantly, these prior studies were not conducted in flowing whole blood, which is a protein-rich non-Newtonian fluid capable of regulating pH through homeostasis. Future work on pH’s effect on coagulation should interrogate the alkaline microenvironment of the material surface with pH microprobes in tandem with live imaging of protein complexes to have a greater understanding of the in vivo impacts.

Currently, evidence suggests a complex and product-rich solution from the degradation of magnesium materials, which is capable of affecting the coagulation response. In this study, corrosion byproducts of magnesium materials, including charged magnesium ions and alkalinization of the device environment, affected the plasma proteins responsible for coagulation. Furthermore, surface modifications like anodization and fluorination significantly increased platelet accumulation and fibrin deposition compared to the base material WE22, indicating an effect of surface modification on the thrombogenic response. Local interrogations of byproduct availability, surface degradation, and protein adhesion are recommended to determine the mechanism of differential thrombus burden between materials. The complex milieu that drives thrombogenesis requires multiple approaches to delineate mechanisms behind reduced thrombogenicity. Balancing the thrombogenicity of a device with its physical characteristics, such as corrosion rate, will be required for advancing to next-generation bioresorbable metal stents. Finally, the biological effects of corrosion byproducts, both detrimental and beneficial to host health outcomes, need to be understood in the context of clinical use of the stenting device.

5. Conclusion

All magnesium alloys without surface treatments prevented thrombus formation similarly to pure magnesium in both whole blood and isolated component environments. In these studies, changing the alloying elements did not change fibrin deposition, platelet accumulation, fibrin clotting time, or FXII activation in a significant way. Surface-modified magnesium materials caused an increase in thrombogenicity ex vivo; however, there were no detectable differences found in vitro. Future work examining the relationship between the approaches of alloying and coating magnesium, which are used to control corrosion rate, and magnesium’s anti-thrombogenic effects, with particular focus on thrombus stability, will aid in the design of more effective and safer bioresorbable metal stents.

Data Availability

Data will be made available on request.