Research ArticleBIOMATERIALS

An injectable shear-thinning biomaterial for endovascular embolization

See allHide authors and affiliations

Science Translational Medicine  16 Nov 2016:
Vol. 8, Issue 365, pp. 365ra156
DOI: 10.1126/scitranslmed.aah5533

Stopping blood in its tracks

Effective treatments for ruptured blood vessels must be rapidly deployed to promote hemostasis. Avery et al. formulated a gelatin and silicate nanoplatelet hydrogel material that occluded blood flow without requiring thrombus formation. When injected into arteries and veins in mice and pigs, the biomaterial occluded blood flow without evidence of fragmentation or displacement for up to 24 days. Occluded vessels showed evidence of connective tissue replacing the biomaterial in the vessel lumen. Shear-thinning biomaterials represent promising alternatives for stable endovascular embolization.


Improved endovascular embolization of vascular conditions can generate better patient outcomes and minimize the need for repeat procedures. However, many embolic materials, such as metallic coils or liquid embolic agents, are associated with limitations and complications such as breakthrough bleeding, coil migration, coil compaction, recanalization, adhesion of the catheter to the embolic agent, or toxicity. Here, we engineered a shear-thinning biomaterial (STB), a nanocomposite hydrogel containing gelatin and silicate nanoplatelets, to function as an embolic agent for endovascular embolization procedures. STBs are injectable through clinical catheters and needles and have hemostatic activity comparable to metallic coils, the current gold standard. In addition, STBs withstand physiological pressures without fragmentation or displacement in elastomeric channels in vitro and in explant vessels ex vivo. In vitro experiments also indicated that STB embolization did not rely on intrinsic thrombosis as coils did for occlusion, suggesting that the biomaterial may be suitable for use in patients on anticoagulation therapy or those with coagulopathy. Using computed tomography imaging, the biomaterial was shown to fully occlude murine and porcine vasculature in vivo and remain at the site of injection without fragmentation or nontarget embolization. Given the advantages of rapid delivery, in vivo stability, and independent occlusion that does not rely on intrinsic thrombosis, STBs offer an alternative gel-based embolic agent with translational potential for endovascular embolization.


Vascular injury causing uncontrolled bleeding is associated with mortality rates of more than 40% (1, 2). Although the traditional open surgical approach has been the cornerstone in the treatment of these patients, there has been a recent shift in the use of minimally invasive endovascular embolization techniques, which can be safer and faster (3). From a pinpoint puncture of the common femoral artery, a combination of long wires and catheters is navigated through the vast arterial system under fluoroscopic guidance to reach the bleeding site for treatment. Once the site of the injured vessel is reached, the standard of medical practice is to embolize these vessels with permanent metallic coils that are pushed out from the catheter; these coils then induce thrombosis to occlude the vessel, preventing further bleeding. Despite its improvements over open surgery, rebleeding after coil embolization continues to be as high as 47% (4). In patients unable to form a thrombus within the coiled artery, such as patients with disseminated intravascular coagulation (DIC), or in patients who are on high-dose anticoagulation for mechanical valves or mechanical cardiac assist devices, breakthrough bleeding can be common, resulting in major morbidity and mortality (5, 6). The incidence of gastrointestinal bleeding is 40% in patients with cardiac assist devices (7). Other complications of coil embolization include coil migration, nontarget embolization, and coil compaction; handling of these coils also requires highly skilled physicians to perform these procedures (810). Follow-up imaging of these patients can pose a limitation because metallic coils produce extensive streak artifact, limiting evaluation of the injured or diseased tissue. These procedures can also be lengthy and costly, exposing the patient and the staff to high radiation doses and limiting access to these costly life-saving interventions to tertiary medical centers (11, 12).

Liquid embolic agents provide an alternative to metallic coils for treating vascular injuries. Onyx, a solution of ethylene vinyl alcohol copolymer in dimethyl sulfoxide, is a liquid embolic agent approved by the U.S. Food and Drug Administration for embolization of intracranial arteriovenous malformations (AVMs) and recently has received a “Humanitarian Use Device” approval for aneurysm embolization (13). However, Onyx is cost-prohibitive and requires extensive preparation and specialized operator training before use; complications are also common, including the dangerous possibility of cementing the catheter to the Onyx released into the AVM or aneurysm sac (10, 14). Furthermore, Onyx has a 20% rate of incomplete aneurysm occlusion (15) and carries a 12 to 36% chance of recanalization (16). Major procedural complications include aneurysm rupture, leakage during injection, angiotoxicity, and the possibility of necrosis (1722).

We aimed to develop an endovascular embolic agent that could (i) be rapidly deployed; (ii) create a complete, impenetrable cast of the bleeding vessel or aneurysm without relying on the patient’s ability to form a clot; (iii) prevent recurrence of bleeding; (iv) allow magnetic resonance imaging (MRI), computed tomography (CT) imaging, or ultrasound (US) imaging without any artifact; and (v) demonstrate use in challenging, high-mortality clinical scenarios, such as in patients with severe intrinsic coagulopathy or those receiving anticoagulation therapy. Shear-thinning or in situ gelling hydrogels are systems that have the potential to address many of these requirements. Their ability to completely fill a space and the variety of gelation mechanisms available could simplify catheter delivery and embolization procedures (23, 24). Previous uses of hydrogels as embolic agents have been described (2427), but many involve long preparatory periods (24), high injection pressures (26), or nondegradable components (25). The lack of an efficacious, safe, and reliable tool for embolization necessitates a more effective technique that is occlusive without relying on thrombosis, less toxic, less costly, less time-consuming, and easier to administer without the need for specialized tools or skills.

Here, we developed and tested in small and large animal models a formulation of a shear-thinning biomaterial (STB), a nanocomposite (NC) composed of synthetic silicate nanoplatelets and gelatin previously investigated by Gaharwar et al. (28), with physical and biological properties suitable for catheter-based endovascular embolization (Fig. 1, A to C). Our previous investigation applied STBs directly over externally injured rat liver, demonstrating its hemostatic potential and rapid recovery after high shear. In this previous study, the STB was also assessed for its rheological behavior and physiological stability. Although the application of the STB on a rat liver surface was hemostatic in nature, the study did not require injectable shear-thinning behavior, and therefore, a concentrated formulation was tested. In contrast, endovascular embolization and other small-bore catheter–based applications require injectability through long catheters (up to 150 cm); instantaneous gelation to create an occlusive seal of the arterial lumen without fragmentation; a formulation that does not cause CT, MRI, or US artifact; and a γ-irradiated material that is sterile and easy to use. These conditions were not previously assessed or characterized in vitro or in small or large animal models, supporting further investigations into the suitability of this STB formulation with optimized properties as an injectable hydrogel embolic agent. Indeed, the STB forms a complete, impenetrable cast of the vessel after injection, which is sufficient to occlude the vessel or the aneurysm without relying on thrombosis. The decrease in viscosity under shear stress, which is an inherent feature of STBs, is expected to simplify delivery through catheters, and the recovery of modulus after injection ensures that the embolic agent remains at the site of injection without fragmentation or nontarget embolization (Fig. 1B). Hemostatic silicate-based nanoplatelets incorporated in the STB have an additional desirable property in that they can promote clotting on the material surface when exposed to blood to enhance thrombosis and further minimize the risk of potential fragmentation.

Fig. 1. STB functions as an endovascular embolic system.

(A) Schematic of STB fabrication. (B) Proposed syringe-based delivery of STB (dark blue) from a catheter (light blue) into a blood vessel to occlude the vasculature and promote local thrombus formation at the exposed ends (yellow). (C) Image highlighting the injectability of the STB through a catheter. Inset is a zoomed image of the STB extruded from the catheter tip. (D) Table with composition information for STBs tested. SN, silicate nanoplatelet; *, viscosity is the maximum value recorded after a 5-min equilibration at 37°C during rheological shear rate sweeps. Data are means ± SD (n = 3). (E and F) Time-dependent plots of shear stress versus shear rate for (E) 6NC50 and (F) 6NC75 immediately (0 min), 5 min, and 60 min after loading, illustrating the shear-thinning properties of the STB (slopes of curves decrease, light blue region). (G) Storage moduli (G′) of 6% (w/v) STBs after repeated application of high strain (100% strain, light blue regions) over time, indicating solid-like behavior.


STB is injectable and stable and quickly recovers modulus

To select STB compositions for investigation as embolic agents, we first performed manual injection assays to determine the highest concentration of silicate nanoplatelets and gelatin that was comfortably injectable through a clinical 5-French catheter. This is important because we consider the ease of use of the STB as a gel-embolic to be a key variable in our design. STBs are composed of a physical mixture of porcine gelatin (type A) and synthetic silicate nanoplatelets (Laponite XLG) (Fig. 1A). When mixed in water, a physically cross-linked hydrogel NC is formed through strong electrostatic interactions. In particular, the negative electrostatic charges on the two sides of the nanoplatelet can interact with positive charges on gelatin chains to generate a stable, homogeneous hydrogel, as previously described (28). STB compositions are identified by the total solid weight (w/v) and the percentage of the solid material consisting of silicate nanoplatelets [for example, 6NC50 is 6% (w/v) and is composed of 50% (w/w) silicate nanoplatelets]. Concentrations ranging from 3 to 9% (w/v) were tested for manual injectability through a 5-French catheter [inner diameter (ID), 0.97 mm]. For further analysis, these injectable compositions of high (>50%) nanoplatelet percentage were chosen to ensure maximum hemostatic activity and maximum solid concentration (6%; Fig. 1D and table S1).

To understand the temporal dependence of STB injectability, which is critical to ensure continuous delivery of STBs throughout the operation window, we monitored the dynamic changes in rheological properties relating to injection over time. Shear-thinning behavior and thermal insensitivity were verified in all STBs before further testing (figs. S1 and S2). Shear rate curves were collected over the span of 1 hour to determine whether there were qualitative changes in their shear-thinning behavior. Shear rate curves at 5 and 60 min (Fig. 1, E and F) after the initial measurement showed no qualitative change in the curve profile, suggesting that shear-thinning behavior of the STB was maintained even after extended periods without applied flow.

To characterize the mechanical stability of the STB after injection through a catheter or needle, we tested the recovery of the modulus after application of high strain (100% oscillatory strain). Previously, 9% (w/v) NCs (28) were shown to rapidly recover solid-like behavior [shear storage modulus (G′) > shear loss modulus (G″)] after application of high strain, which resulted in liquid-like behavior in the NC (G″ > G′). Similar experiments using 6% (w/v) STBs indicated comparable recovery after application of high strain (Fig. 1G and fig. S3).

Next, to replicate the injection process, we investigated STB injectability through clinical catheters and needles. To this end, the force required to inject the STB from 3-ml syringes through clinical catheters or needles at a set flow rate was measured with a mechanical tester (Fig. 2A and fig. S4). Parameters, including the material composition, injection rate, and catheter or needle dimensions, affected the force profile during injection (fig. S5). The applied force linearly increased in all tested catheters and needles until it plateaued at the injection force (Fig. 2B); this was achieved once the STB began to extrude from the catheter or needle tip (movie S1). These injection forces are within a range that can be generated manually without the aid of additional equipment, as supported by table S1. Additionally, the final injection force was similar whether the catheter was filled with STB or empty before measuring the injection force once flow was fully developed (fig. S5), indicating that shear-thinning behavior was unaffected by the presence of STB remaining in the catheter.

Fig. 2. STB is injectable through clinical catheters and is occlusive in in vitro model.

(A) Injection force measurement setup using an Instron mechanical tester. (B) Force required to pass STB through catheter. The average value of the plateau, indicated by the arrow, is used to quantify the injection force. (C) Injection force dependence on injection flow rate. ns, not significant. (D) Table of catheter and needle dimensions (L, length; G, gauge; F, French) used for testing injection force. (E) Injection force of STB through needles and catheters. (F) Injection force for different materials. (G) Schematic of polydimethylsiloxane (PDMS) tube–based in vitro occlusion setup using a syringe pump and a pressure gauge (P) to assess the pressure required to push anticoagulated blood through the PDMS tube containing coils or STB compared to blood alone (control). (H) Pressure curves over time illustrating the control (blood alone), coil, and STB samples (6NC75). (I) Mean pressure required to displace control (blood alone), coil, and STB samples (6NC75). Data are means ± SD (n = 3) for (C), (E), (F), and (I). P values determined by one-way analysis of variance (ANOVA) with Tukey post hoc comparisons for (C), (E), (F), and (I) [not significant (ns), P > 0.05].

To assess the impact of injection parameters on the injection force and to provide comparisons to standard materials delivered through catheters, we varied catheter dimensions, needle sizes, and the rate of embolic delivery while measuring the injection force. Higher injection rates required higher injection forces (Fig. 2C), as expected from Poiseuille’s law for fluid flow in tubes (29). When catheters with different IDs were compared (Fig. 2, D and E), the clinical 4-French (ID, 0.89 mm) and 5-French (ID, 0.97 mm) catheters required similar injection forces (27 ± 0.6 and 24 ± 3.1 N, respectively) (Fig. 2E). However, catheter length was not equal for the 4-French (70 cm) and 5-French (100 cm) clinical catheters. Correcting for the longer length of the 5-French catheter, according to Becker et al. (26), the dependence of injection force on diameter was more apparent (fig. S5D). Common medical solutions delivered by catheters were also measured, including contrast dye (Visipaque) and saline solution (Fig. 2F). The force required to inject STBs was one order of magnitude higher than the force needed to inject saline solution and contrast dye but was still injectable by hand (table S1). These results supported further investigation of STBs’ potential as an embolic agent.

STB effectively occludes vasculature and promotes coagulation in vitro

To evaluate the ability of the STB to withstand physiological pressures without fragmentation, we established an in vitro model to monitor the pressure required to eject the STB from a polydimethylsiloxane (PDMS) tube (diameter, 4 mm) mimicking a vessel (30). The model was established with a syringe pump to flow anticoagulated blood (citrated) through the PDMS tube while the pressure upstream of the tube was being monitored (Fig. 2G). The PDMS tube was attached to the end of the tubing connecting the syringe pump and pressure gauge, and flow rates of anticoagulated blood through the patent channel were tuned to achieve a pressure reading of ~16 kPa, equivalent in SI units to the standard reported systolic blood pressure of 120 mmHg (fig. S6) (31).

Delivery of commonly used Interlock coils (diameter, 4.5 mm; Boston Scientific) or 6NC75 STB into the tubing upstream of the channel was performed by passing a catheter through the tubing connecting the syringe pump to the channel and delivering the embolic agent. When flow started, a significant (P < 0.0001, compared to coils) increase in pressure was observed among STB-containing systems (Fig. 2H), whereas coils alone resulted in pressures within the same range as the blood-only control pressure (16 kPa) (Fig. 2I). The pressure obtained from the STB-only occluded channel (31.7 ± 1.9 kPa) was almost twice the systolic blood pressure (16 kPa) and 24 times the venous pressure (1.3 kPa) (31), suggesting its stability at physiologically relevant pressures in anticoagulated blood. Comparable STB stability and displacement pressures were also observed in ex vivo occlusion studies with explanted porcine vessels (fig. S6) and rheological creep studies applying the magnitude of shear stresses encountered along a vasculature wall (fig. S7). This highlights the mechanical stability of 6NC75 and its ability to physically occlude vessels in the absence of thrombus formation.

To compare the hemostatic activity and hemocompatibility of STBs with commercially used coils, we performed in vitro coagulation and hemolysis tests. The surface of the STB exhibited similar activity as clinical metallic coils (Fig. 3, A and B), the standard treatment for endovascular embolization. Clotting time and hemolysis assays of whole blood in contact with either the STB or metallic coils were performed to assess the hemocompatibility of the STB compared to metallic coils. Clotting times were accelerated when blood was in contact with either coils or the STB, compared to controls (Fig. 3A and fig. S8), as assessed by qualitative analysis. Clotting times between 3 and 5 min were observed for 6% (w/v) STBs, similar to clotting times obtained from coils. Further comparisons of coils and the STB showed higher values of hemolysis for the biomaterial compared to coils (Fig. 3C and fig. S9). However, hemolysis values of STBs were comparable to other engineered hemostats (32). The potential for STB degradation was investigated by incubation with plasma, with no degradation observed in the first 24 hours (Fig. 3D).

Fig. 3. STB is hemocompatible in vitro.

(A) Plate assay (96-well) of blood clotting in contact with commercially available metallic coils, STB, and control (polystyrene substrate). (B) Images of clinical embolic coils (Cashmere and Tornado) and STB. (C) Analysis of red blood cell hemolysis in contact with STB. Control is red blood cells incubated in saline (0.9% NaCl). (D) Degradation of 6% (w/v) STB at 37°C incubated in plasma. (E) An in vitro embolization model used to simulate the in vivo vascular occlusive potential of STB. Grayscale images illustrate higher magnification views of inset from a representative occlusion study. (F) Mean decrease in clotting time of blood in contact with STB was not significantly (P = 0.7857) changed by STB γ-irradiation sterilization and addition of contrast dye (Visipaque; γ + contrast). (G) STB injection force was comparable before and after sterilization. Data are means ± SD (n = 3). P values determined by one-way ANOVA with Tukey post hoc comparisons for (C), Kolmogorov-Smirnov test for (F), and one-way ANOVA with Tukey post hoc comparison for (G) (P > 0.05).

STB forms casts of vasculature and is stable and functional after sterilization

To highlight the ability of the STB to conform to complex lumen contours, we performed additional in vitro occlusion studies in vascular embolization models (Fig. 3E and fig. S10). To visualize the STB by fluoroscopy imaging, we mixed the material with Visipaque, a clinical intravascular iodine-based contrast dye. As shown in Fig. 3E, the STB was manually delivered in a preloaded syringe through a 5-French catheter. The biomaterial could be continuously injected by hand and completely occlude the vessel, creating a cast (movie S2). Contrast dye was observed to remain within the STB and not leach from the biomaterial, suggesting its stability within the STB and its potential to function effectively in vivo. In preparation for in vivo studies, sterilization of the STB with γ-irradiation was also performed to eliminate any infectious pathogens (figs. S11 and S12); γ-irradiation had no statistically significant impact on the clotting ability of the STB (P = 0.7857) (Fig. 3F) and resulted in similar STB injection forces (Fig. 3G). The preserved activity of STBs after sterilization allows for sterile syringes to be loaded before the procedure and to function as previously characterized.

In vivo injection of STB into vessels embolizes vasculature

To assess STB for the possibility of nontarget embolization in vivo, we injected STBs into mouse femoral arteries initially before testing in porcine models. 6NC75 was injected directly into femoral arteries (Fig. 4A), showing sustained and complete occlusion of the vessel that resulted in interruption of hindlimb perfusion on laser Doppler imaging (Fig. 4B). There was no sign of recanalization of the embolized vessel after occlusion and no detectable nontarget distal migration by microCT (SkyScan1275, Bruker) imaging to visualize the Visipaque-labeled STB (Fig. 4C); these data indicated the generation of a stable occlusion without fragmentation, consistent with ex vivo experiments in Fig. 2H. Optimal concentrations of Visipaque contrast agent mixed with STB were determined based on phantom image analysis (fig. S13).

Fig. 4. Arterial injection of STB in mice occludes vasculature.

(A) Photographs of a mouse femoral artery before and after injection of STB through a 30-gauge needle (n = 4). Arrows indicate the approximate location of the injection. (B) Laser Doppler microperfusion imaging showing that hindlimb perfusion was maintained before STB injection (left panel) but was interrupted in the STB-injected hindlimb (right panel, arrow). (C) μCT image of mouse after STB (arrow indicates 6NC75 with Visipaque; artificially colored blue) injection showing the injected radiodense biomaterial stable in the femoral artery, consistent with ex vivo experiments (fig. S6).

To demonstrate feasibility and applicability for potential future clinical use, we used porcine models to deliver 6NC75 STB to target vessels through standard clinical percutaneous catheterization techniques and tools. Because the fragmentation and displacement of the STB after embolization of a target vessel is a critical undesired outcome, initial experiments were designed to assess stability of the injected STB in the arterial vasculature (Fig. 5, A to F). Female Yorkshire pigs were anesthetized, intubated, and monitored throughout the procedure. Under US guidance, the right carotid artery was accessed using a standard 21-gauge needle and wire. Under fluoroscopic guidance, a 5-French Cobra 2 catheter (Cook Medical) and a 0.035″ angled Glidewire guidewire (Terumo) were brought to the infrarenal aorta, and digital subtraction angiography (DSA) was performed, demonstrating the lumbar arteries and the iliac arteries. The lumbar artery was catheterized and successfully embolized within seconds after the injection of about 1.5 ml of γ-irradiated 6NC75 (Fig. 5, A to C). Figure 5C shows a magnified view of the STB edge with patent proximal lumbar artery and aorta. Because this STB remained in place and did not migrate over a period of 15 min, we next embolized the left external iliac artery (EIA). The same catheter was brought just distal to the internal iliac artery origin, and about 4 ml of the STB was injected (Fig. 5, D to F). Flow immediately ceased without any evidence of displacement or fragmentation, indicating that the STB remained at the injected site. This was an important outcome because EIA is at high flow (about 100 cm/s), with a diameter of 5 to 7 mm, similar to the human iliac arteries (10 to 12 mm) (33, 34); when compared to the clinical scenario, coil embolization of this high-flow, large-diameter artery would be at high risk for coil displacement, which was not the case for the STB.

Fig. 5. Vascular embolization in porcine model suggests feasibility.

(A) Angiography of the normal right L4 lumbar artery in female Yorkshire pigs (n = 4). (B) Five-French Cobra 2 catheter within the right L4 lumbar artery after STB injection; contrast injection revealed lack of opacification of the lumbar artery, indicating successful embolization. (C) Magnified view showing the tip of the catheter inside the lumbar artery and the abrupt cutoff of the artery, indicating an impenetrable cast of the vessel and nonopacification despite high-volume, high-velocity contrast injection (>1000 psi). (D) DSA at the level of the aortic bifurcation. (E) Abrupt cutoff of the left EIA after STB injection. (F) Magnified view of the embolized EIA. (G) Embolization of various forelimb central veins using a 5-French catheter. Arrows indicate abrupt cutoff of veins after embolization. (H) Coronal CT study at 24 days after injection of STB. (I) CT iodine map demonstrating clear lungs without any pulmonary embolism 24 days after injection. White arrows in (G) (middle image), (H), and (I) indicate the same regions of embolized vein. (J) Gross image revealing the distended, occluded vein in (G) (middle image). (K) The cut surface of the STB occluded vein after 24 days showing STB (arrow) within the vein.

Next, we explored the durability of STB embolization in the forelimb venous vasculature. Forelimb veins were chosen because any fragmentation or displacement of the STB would result in pulmonary embolism, an objective assessment of performance that was tested using CT imaging. Using a 5-French Davis catheter (Cook Medical), various centrally positioned forelimb veins were embolized in three pigs using 3 to 4 ml of the γ-sterilized 6NC75 STB (Fig. 5G). These pigs survived for 14, 18, and 24 days; just before necropsy, CT imaging was performed using a 256-slice dual-energy scanner (Definition Flash, Siemens) before and after intravenous administration of iodinated contrast medium (Visipaque). CT imaging revealed no evidence of artifact from the STB that would obscure the region of embolization. CT imaging also revealed that there was no evidence of pulmonary embolism, as indicated by the highly sensitive iodine maps of the lungs up to 24 days after embolization (Fig. 5, H and I). After CT imaging, the embolized veins were carefully dissected and immediately processed for histologic and immunohistochemical assessment. On gross examination, the veins were intact and occluded, with the STB completely filling the vein at the initial site of injection with no signs of recanalization, even after 24 days (Fig. 5, J and K).

Transverse histology sections obtained from the embolized vessels in Fig. 5 were analyzed, showing intraluminal obliteration and remodeling of the vessel with fibrous connective tissue when compared to control patent vessels (Fig. 6). Paraffin-embedded cross sections were obtained from the proximal aspect to the distal portion of the embolized vessel segment and were stained with hematoxylin and eosin (H&E) and Masson’s trichrome (fig. S14). Trichrome-stained sections indicated that the tissue was rich in connective tissues (collagen stained blue). Residual STB was present in the vessel with infiltrating inflammatory cells evident in the vessel lumen surrounding the remaining biomaterial. Immunohistochemistry for myeloperoxidase (MPO) revealed MPO-positive cells infiltrating the vessel luminal area. MPO is expressed primarily by neutrophils of the myeloid cell lineage but can also be present in monocytes and macrophages (35). There was negligible MPO staining observed in the vessel wall, highlighting that inflammation was confined to the lumen (Fig. 6A, first row). Macrophages, as indicated by CD68 staining, were predominately expressed surrounding the remaining STB (Fig. 6A, second row), suggesting that there was ongoing degradation and clearance of residual STB inside the occluded vessel by phagocytic cells. In addition, proliferating cell nuclear antigen (PCNA) staining indicated that there was active cellular proliferative activity (Fig. 6A, third row). The PCNA staining localized primarily to the areas of residual STB, suggesting that there is continued remodeling that parallels STB clearance within the occluded vessel lumen.

Fig. 6. Histological assessment of vascular reorganization after STB injection.

(A) Representative histology staining results from patent and occluded veins after STB injection in a porcine model (n = 4). In occluded vessels, histology sections of MPO, CD68, and PCNA showed myeloid cells, macrophages, and proliferating cells surrounding the remaining STB (asterisks) with staining localized to the lumen. (B) Replacement of STB with connective tissue in the vessel was quantified by measuring the amount of intraluminal area filled with connective tissue from H&E histology images over time, reflecting the percentage of vessel remodeled (n = 6). (C) Vascular remodeling (%) as a function of location proximal or distal to STB injection (n = 8). Location relative to STB injection (x axis) corresponds to (D). (D) Histology slides indicating spatial dependence of vessel remodeling according to locations more proximal or distal within the site of embolization. Unstained areas are artifacts of the histology process and represent areas of the vessel still filled with STB.

To better understand the impact of STB on vessel remodeling, we performed morphometric quantification of the occluded vessel area. Clearance of the STB and remodeling of the embolized vessel were dependent on temporal and spatial factors (Fig. 6, B to D). Temporally, histological sections from later time points were more advanced in their clearance of STB and subsequent vessel remodeling. An increase in the percent of the vessel remodeled with connective tissue was observed as the STB was cleared from the site (Fig. 6B). Spatially, comparing proximal and distal histologic cross sections of the embolized vessel, a trend toward decreased connective tissue presence was noted. There was near 100% connective tissue occupying the vessel at proximal locations and a decreased presence of connective tissue in more distal sections (Fig. 6, C and D). The most distal section showed an increase in the presence of connective tissues relative to its adjacent more proximal sections, likely because of its proximity to flowing blood and the ability for phagocytic cells to interact with the STB at this distalmost location during remodeling. The high percentage of connective tissues present at both the proximal and distal sites of occlusion resulted in an embolized site with minimal likelihood for subsequent fragmentation or recanalization.


Minimally invasive interventions are desired to reduce morbidity and mortality associated with open surgical procedures and to improve overall patient outcomes (36). For minimally invasive procedures that involve drug delivery (37, 38) or placement of devices (39), catheters are an important tool to provide access to otherwise inaccessible locations in the vasculature. In particular, endovascular embolization procedures can deliver embolic agents within the vasculature to occlude bleeding vessels or aneurysms (10, 40).

Rapid and controlled delivery of embolic agents is required for successful embolization procedures. Coiling is the most common procedure used to embolize vascular injuries (41, 42), with similar performance in bare and hydrogel-coated coils (43). However, the distance from the site of insertion in the groin to the site of vascular injury, tortuous vasculature, and the size of the target vessel can hinder coil delivery and impact clinical outcomes. Coil embolization solely relies on intrinsic thrombosis and is therefore not suitable for patients with coagulopathy or on anticoagulation therapy. Liquid embolic agents and hydrogels have simplified delivery through tortuous pathways and do not require blood clotting (10). The liquid embolic agent Onyx has been predominately used in intracranial arteriovenous malformation treatment (44). In situ gelling agents, such as cyanoacrylate, alginate (26, 45), poly(ethylene oxide)/poly(propylene oxide), and other thermoresponsive polymers (27, 46) have also been investigated as alternatives. However, time-sensitive gelation in Onyx and cyanoacrylate-based embolic agents, complications such as cementing of the catheter to the embolic material or nontarget embolization and artifacts from cross-sectional imaging as well as cost limit their use.

With the STB described here, we have generated a promising gel-based embolic agent that can be rapidly deployed from preloaded sterile syringes, completely occludes targeted arteries and veins, shows no indications of recanalization after injection, and is capable of performance in anticoagulated blood or in coagulopathic scenarios. During injection, the STB in the bulk does not enter a liquid state, likely exhibiting a plug flow process that allows the central core of the hydrogel to pass through the catheter or syringe because the hydrogel along the wall is fluid and acts as a lubricant, achieving a rapid recovery time after encountering high stress or strain (47). Therefore, the injectability is time-insensitive within the period of standard procedure, providing freedom for physicians to deliver the embolic without the threat of catheter blockage or potential adhesion to the embolic material and eliminating the need for specialized delivery tools. The hydrogel enables facile navigation of tortuous paths for delivery. After placement in the vasculature, we have shown in vitro and in vivo that the injected STB is stable and capable of withstanding physiological pressures without being displaced.

Catheter blockage was not observed during STB injection, even when the STB was allowed to remain immobile in the catheter. Using rheological analysis, we demonstrated that the shear-thinning properties of the biomaterial could be maintained for at least 1 hour. This suggests that prolonged procedures can also be accommodated without risking catheter blockage, an advantage for physicians dealing with complicated cases that could require repeated angiography imaging of the vasculature during injection. In contrast, hydrogel-coated coils swell, limiting precise delivery and preventing repositioning (10, 48). The curve profile observed in the rheological shear stress and mechanical tester measurements of injection force offer a glimpse into the yielding behavior of different STB compositions, possibly yielding through shear banding and plug flow process (47). The gradual curvature in the injection force curves seen in rheological and catheter injection of 6NC50 suggests a more controlled injection process in contrast to the sharp break in curvature seen with 6NC75 injection. Control over the injection profile can provide better user control during injection, allowing for smooth delivery of the embolic material. Creep compliance or the tendency of the gel to deform under loading was observed in the STB but, at the relevant shear stresses encountered in vivo (49), are less than 1 Pa. Almost complete recovery after shear was observed as well, suggesting minimal permanent deformation in the STB. Generation of a clot surrounding the hydrogel and eventual remodeling into dense fibrous connective tissues will further stabilize the hydrogel and prevent excessive creep for a long term.

Incomplete occlusion is not a concern with STB because it can fully occlude the vasculature without the requirement of thrombus formation, which is necessary in coiling procedures (50). The STB is capable of higher packing densities than bare coils, hydrogel coils, or other alginate-based embolic agents because it creates a complete cast of the vessel wall, decreasing the likelihood of recanalization due to insufficient packing (45, 50, 51). Additionally, no imaging artifacts were noted when imaging the STB. This is an important outcome because streak artifact from metallic coils limits noninvasive assessment of the embolized segment; for example, coil-embolized aortic or cerebral aneurysms make CT assessment for an endoleak nearly impossible, necessitating an additional angiographic procedure for the patient (52). The intrinsic procoagulant property of the STB accelerates the natural coagulation that will occur on its surface, further stabilizing the occlusion and preventing fragmentation or nontarget embolization. Coagulation occurs only on the exposed surface of the STB compared to metallic coils where coagulation is initiated at the interface of the coil with blood. However, even when bleeding has progressed into coagulopathic conditions because of reduced clotting factors, physical occlusion of the bleeding vasculature with the STBs can still assist in controlling bleeding, showing clear advantage over the coils that solely rely on thrombosis. As the STB is removed by macrophages or other cell types, proliferating cells deposit connective tissue in the vacated area, ensuring a stable, chronic occlusion. The ability of the body to remodel and remove the STB presents an opportunity to introduce stimuli to promote a variety of activities in vivo, including control over further angiogenesis (53) with antiproliferative drugs (54).

Although coagulation can improve STB stability in the occluded vessel along the blood-exposed edges, the STB is limited in its ability to strongly adhere to the vessel tissue, permitting the unwanted mobilization of the STB plug under supraphysiological pressures or large deformations of the occluded vessel. This is an intrinsic property of shear-thinning materials, which will decrease in viscosity under high shear unless reinforced by some orthogonal network in vivo (55). Cell adhesion can be increased in biomaterials by using cell-adhesive peptide sequences (56) or fibrin mimetic polymers (57), improving the interaction of the STB with the natural clot and surrounding vessel tissue while maintaining its injectability and procoagulatory properties.

Shear-thinning materials and other in situ gelling hydrogels represent a viable, safe, and cost-effective alternative to coil embolic systems, compatible with the minimally invasive procedures that are more common for endovascular embolization. Coupling easy deliverability with stability after injection and complete vessel occlusion are required for any embolic material. This study sets the stage for future studies and clinical trials to investigate the potential of shear-thinning, in situ gelling hydrogels in the management of vascular injuries or abnormalities and other conditions requiring endovascular embolization.


Study design

The study was designed to determine the optimum STB formulation for endovascular embolization. It was hypothesized that the highest concentration material with the highest nanoplatelet concentration that was also injectable through a catheter would provide the best combination of deliverability, procoagulatory activity, and stability after delivery. After characterization and optimization in vitro and ex vivo, we examined whether the STB was capable of embolization in vivo in mice (n = 4) and porcine models (n = 4). Power analysis indicated that n = 4 would detect an effect size of 2 with 80% power and α = 0.05. STB was injected into arteries in mice to assess its ability to occlude flow in small animals in the short term, and STB was injected into the vasculature of swine to assess the long-term feasibility of embolization in larger animals. Arteries and veins were embolized through catheter delivery. Time points (days 14, 21, and 24) were chosen to allow for remodeling to occur in the embolized veins, to assess them by histology, and to determine the stability of the STB and its impact on tissue deposition. Data were not blinded. No data were excluded from this study.

STB formation

Stock solutions of gelatin [18% (w/v); G1890, Sigma] and silicate nanoplatelets [9% (w/v); Laponite XLG, BYK] were made in Milli-Q water and mixed to generate NC gels (28). Solid gels were generated from all the tested compositions.

Rheological analysis of STBs

STB shear rate sweeps and recoverability assays were analyzed according to protocols previously developed (28).

Injection force of STB

The injectability of the material was analyzed using a mechanical tester (Instron Model 5542). Briefly, the STB was added to 3-ml plastic syringes (ID, 8.66 mm; BD Biosciences) and injected through medical catheters (4-French Beacon, pediatric pigtail flush, and 5-French Beacon, multipurpose A; Cook Incorporated) or needles (18 gauge and 23 gauge, BD Biosciences) using standard luer lock fittings. The syringe plunger was depressed using an upper compressive platen, and the housing of the syringe or catheter was fitted into a lower tensile grip to prevent movement during the experiment. The injection rate was controlled by changing the cross speed of the compression platen to achieve the desired flow rates. The force on the plunger was measured with a 100-N load cell and recorded using Bluehill version 3 software.


Citrated human whole blood (Research Blood Components) was used for all blood-related testing.

Clotting time. Blood was mixed with 10% (w/v) 0.1 M CaCl2 in Milli-Q deionized (DI) water and pipetted to mix. Clotting times were measured according to previously developed protocols (28).

Degradation. Human plasma was separated from citrated whole blood by gravity-settling of the red blood cells. Plasma was used to incubate weighed STB (500 μl) in 1.5-ml Eppendorf tubes (VWR) at 37°C. At time points during the first 24 hours, the plasma was removed, the remaining STB weighed, and replaced with fresh plasma.

Hemolysis. Hemolysis testing was performed according to protocols from Kumar et al. (32). Briefly, citrated whole blood was diluted 50× into 0.9% (w/v) saline solutions. STB was flattened in 1.5-ml Eppendorf tubes in a swinging bucket centrifuge. Equal volumes of diluted blood and either STB, saline (negative control), or DI water (positive control) were incubated at 37°C for 2 hours under agitation in a shaker incubator (100 rpm, Labline Instruments). Samples were centrifuged (2000 rpm, Labnet), and the supernatant was transferred into wells of a 96-well plate. Percent hemolysis was calculated according to the following equationEmbedded Imagewhere Asample is the absorbance at 545 nm of the STB-containing supernatant, Aneg is the absorbance of the saline diluted blood, and Apos is the absorbance of the DI water diluted blood.

In vitro occlusion

PDMS tubes (ID, 4 mm; wall modulus, about 500 kPa) were fabricated to mimic blood vessels according to our recently developed method (30). Anticoagulated whole blood (citrated, Research Blood Components) was flowed through the tube with a syringe pump, with the pressure being monitored upstream. A 5-French (ID, 1.7 mm) catheter was fed through the tubing to inject coils or STB directly into the PDMS vessel mimic. The catheter was removed, flow was started, and the pressure was measured until movement of the embolic agents in the tube.

In vivo studies

All animal experiments were performed in accordance with the U.S. Animal Welfare Act, following institutional guidelines and animal use protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the Massachusetts General Hospital and according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

In vivo mouse model

STBs were mixed with contrast dye (Visipaque) at a ratio of 1:30 Visipaque/water. Sterilization was performed by γ-irradiation (see the Supplementary Materials). After induction of anesthesia, mice (12-week-old male C57BL6 mice, n = 4) were placed in a supine position over a warming platform to maintain a core temperature of 37°C. Laser Doppler perfusion imaging of the hindlimbs was performed to obtain per-injection baseline scans. A 0.5- to 1-cm incision was made longitudinally on the anterior thigh of one hindlimb. The vessels in the thigh were exposed under a surgical microscope with a combination of sharp and blunt dissection to expose the femoral artery and vein. The femoral artery segment distal to the inguinal ligament was mobilized and freed of its surrounding tissue. Two 6-0 silk sutures were passed distally and proximally underneath the artery to allow for gentle manipulation of the artery by lifting and aligning the vessel for injection. The artery was injected distally with STB (6NC75) using 30-gauge sterile syringe needles. After injection, the incision was closed with a 5-0 polypropylene suture, and the mouse remained on the warming table for 1 hour under general anesthesia followed by post–STB injection laser Doppler perfusion imaging. The perfusion scanning of the injected hindlimb was compared to that of the noninjected contralateral hindlimb to confirm ischemia. After laser Doppler imaging was completed, animals were euthanized, and the carcasses were placed inside a nonopaque container for contrast-enhanced microCT scanning imaging.

MicroCT imaging. MicroCT was completed to produce high-resolution three-dimensional (3D) images constructed of 2D transaxial projections of the euthanized mouse hindlimb containing the radiodense contrast–containing STB. Imaging of the vasculature was completed 1 hour after injection using the Nikon XT H 225 (microCT) and Bruker SkyScan 1275 imaging scanners. The parameters used in the Nikon platform were 70 kV, 20 W, 2400 projections, 4 frames per projection, and 225-ms exposure; for the SkyScan system, the parameters were 70 kV, 166 μA, 2400 projections, and 4 frames per projection. The Nikon specimens were reconstructed with Nikon’s CT Agent and CT-Pro 3D software and viewed in VGStudio MAX version 2.2. The SkyScan specimens were reconstructed with SkyScan’s NRecon software and viewed in their CTAN version 1.15 software.

In vivo porcine model

After approval of the study protocol by the IACUC of the Massachusetts General Hospital, female Yorkshire swine (Sus scrofa domestica; n = 4; weight, 50 to 55 kg; Cummings School of Veterinary Medicine at Tufts University) were purchased. The swine were allowed to acclimate for at least 2 days in a cage; the night before procedure, food was withheld, but water was provided ad libitum. For the procedure, all swine received tiletamine-zolazepam (5 mg/kg intramuscularly; Telazol, Zoetis) and atropine (0.04 mg/kg intramuscularly) as preinduction medication and were then placed supine on the operating table. After anesthesia induction (20 mg/kg intravenously; propofol), all swine were intubated (ID of endotracheal tube, 7.5 to 8.0 mm) and ventilated mechanically (Evita 4, Dräger).

During the procedure, electrocardiogram, transcutaneous oxyhemoglobin saturation (SpO2), end-tidal CO2 concentration, inspired oxygen fraction, and core temperature were monitored. A bolus of propofol (1 mg/kg intravenously) was used to maintain general anesthesia. After induction of anesthesia, access to the carotid artery or to the common femoral vein was obtained using US and C-arm fluoroscopy (Siemens). The access needle and wire were exchanged for a 5-French catheter (Cook Medical). Using a Glidewire (Terumo), the target vessel was selected and embolized using sterilized STB (6NC75), as described above. After the procedure, the catheter was removed, and hemostasis at the puncture site was obtained using manual compression for up to 15 min. The animals were subsequently recovered and placed back into their cages.

Statistical analysis

Normality was determined with the Shapiro-Wilk test. For data sets that were not normal, the nonparametric Kolmogorov-Smirnov test was used for two group data sets, and the Kruskal-Wallis method with Dunn’s multiple comparisons test was used for ≥3 group data sets. For data sets that were normal, Student’s t test was used for two group data sets, and one-way ANOVA with Tukey post hoc tests were performed for ≥3 group data sets. P < 0.05 was defined as significant for all statistical tests. SD was the measure of uncertainty in all data. All statistical analysis and graphing were performed with the GraphPad Prism version 7 software.


Materials and Methods

Fig. S1. Model shear-thinning.

Fig. S2. STB thermal stability.

Fig. S3. Zoomed graph of rapid STB recoverability.

Fig. S4. STB injectability setup.

Fig. S5. STB injection forces.

Fig. S6. Ex vivo occlusion.

Fig. S7. Creep measurement of STB.

Fig. S8. STB thrombosis potential.

Fig. S9. Hemolysis potential of embolic coils.

Fig. S10. In vitro embolization model.

Fig. S11. Impact of sterilization on STB.

Fig. S12. STB sterility by colony formation count.

Fig. S13. Imaging phantom of STB.

Fig. S14. Histology staining of STB occluded and patent vessels.

Table S1. Preliminary injectability.

Movie S1. Injection of STB through catheter.

Movie S2. CT imaging of STB in vascular model.

References (5860)


  1. Acknowledgments: We thank M. Griffin for performing the μCT scans. We also thank X. Li for providing technical assistance. We appreciate S. M. Moosavi-Basri for assistance with illustrations in Fig. 2G. We thank B. Joughin for assistance with statistical tests. Funding: R.K.A. was supported by the NIH Interdepartmental Biotechnology Training Program (NIH/NIGMS 5T32GM008334). This research was supported by the U.S. Army Research Office under contract W911NF-13-D-0001. Y.S.Z. acknowledges the National Cancer Institute of the NIH Pathway to Independence Award (1K99CA201603-01A1). A.K. acknowledges funding from the NSF (EFRI-1240443), IMMODGEL (602694), and the NIH (EB012597, AR057837, DE021468, HL099073, AI105024, and AR063745). R.O. acknowledges funding from the NIH (EB021148 and CA172738) and the Mayo Clinic. Author contributions: A.K. and R.O. generated the concept. R.K.A., M.A., Y.S.Z., and R.O. designed and R.K.A., M.A., and Y.S.Z. performed in vitro and ex vivo experiments. R.O. and M.J.D. performed all porcine in vivo experiments, and R.O. and H.A. performed all murine in vivo experiments. R.K.A., M.A., and R.O. wrote the manuscript. H.A., Y.S.Z., R.O., B.D.O., and A.K. analyzed the experiments, and H.A., Y.S.Z., B.D.O., and A.K. provided comments to the manuscript. D.V.S. performed the CT scans. Competing interests: R.K.A., A.K., and B.D.O. are inventors on a patent application (61/836,761) titled “Nanocomposite hydrogels” submitted by Brigham and Women’s Hospital (BWH) that covers the composition of the STB. R.O. and A.K. are inventors on a patent application (62/194,644) titled “Shear-thinning compositions as an intravascular embolic agent” submitted by BWH that covers application of the STB. Data and materials availability: All data are included in this paper or the Supplementary Materials. Additional materials can be obtained by request to A.K. and R.O. and may be subject to material-transfer agreement requirements with BWH.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article