Bicomponent nano- and microfiber aerogels for effective management of junctional hemorrhage

Fabrication of biocomponent aerogels

The bicomponent aerogels were fabricated through a series of processes, including electrospinning, wet electrospinning, freeze-casting, and cross-linking²². The initial step involved creating short NFs from an NF mat. Specifically, 2 g of PLA (MW: 60 kDa) pellets (Sigma-Aldrich, St. Louis, MO, USA) were dissolved in a 20 mL solution consisting of 16 mL dichloromethane (DCM) (Oakwood Chemical, Estill, SC, USA) and 4 mL dimethylformamide (DMF) (Oakwood Chemical). To enhance hydrophilicity, 0.01 g of Pluronic F-127 (Sigma-Aldrich) was added to the PLA solution. This solution was then loaded into a 20 mL syringe for electrospinning. The electrospinning process was carried out at a flow rate of 0.6–0.9 mL/h using a syringe pump (Fisher Scientific, Pittsburgh, PA, USA) under a voltage of 15–18 kV with a 21-gauge needle (the spinneret) connected to a high voltage generator (ES304-5W/DAM, Gamma High Voltage Research Inc, Ormond Beach, FL, USA). Nanofibers were collected on a grounded mandrel, which was 10 cm long, 12 cm in diameter, and rotating at 1200 rpm, positioned 13 cm from the needle. After collection, the NF mat was removed from the collector, air-dried for 24 h, and then treated with air plasma (PDC-001-HP, 115 V, Harrick Plasma, New York, USA). The fibers were cut into lengths of 60 µm, 30 µm, and 20 µm using a cryostat, then dispersed in water to create suspensions, freeze-dried, and stored at 4 °C for further use.

The subsequent step involved the fabrication of short MFs using wet spinning. To prepare a 20% w/v PCL (MW: 80 kDa) (Sigma-Aldrich) solution, 2 g PCL pellets were dissolved in 10 mL mixture of DCM and DMF, with each solvent contributing 5 mL. This solution was then loaded into a 20 mL syringe and extruded at a rate of 3.0 mL/h through a 3D printed device equipped four 21-gauge needles. The extruded solution was directed into a coagulation bath containing 200 proof ethanol (Decon Lab, King of Prussia, PA, USA). The extrusion device, designed with four outlets and one inlet, was created using a Vida 3D printer (EnvisionTEC, Gladbeck, Germany) and fabricated from Clear Guide material (EnvisionTEC). A low-speed plastic mandrel wrapped in aluminum foil, positioned above the ethanol bath, was used to collect the solidified MFs. After the wet spinning process, the MF bundle was thoroughly dried for several days, cryostat cut under liquid nitrogen to prevent pressure-induced fusion, and treated with air plasma using a High Power Expanded Plasma Cleaner (PDC-001-HP model, 115 V, Harrick Plasma, New York, USA).

Next, short NFs, MFs, or at 50/50 w/w combination of both (NF/MF) were dispersed in water to form suspensions. These dispersed fibers were then homogenized to prepare using a probe homogenizer set at 20% amplitude, with 20/10 s on/off cycles, for 2.5 h under ice-cold conditions. Following homogenizations, the short NF, MF, or NF/MF suspensions were mixed with 1% gelatin granules (derived from bovine skin, Type B, Sigma-Aldrich), and further homogenized for another 2.5 h under the same conditions. The resulting homogenized suspension was then poured into a copper mold attached to an aluminum plate and quickly transferred to a −80 °C freezer, where it was left for 12 h. After freeze-drying, the samples were cross-linked by exposure to glutaraldehyde vapor (EM grade, 2.5% in anhydrous ethanol, Ladd Research, Cincinnati, OH, USA) for 6, 12, or 18 h. NA and MA were created using the same method as NMA. The key difference is that NA lacks MFs, while MA lacks NFs. Finally, the samples were sterilized using ethylene oxide gas (Anprolene AN7916 Ethylene Oxide Ampoules, Andersen Sterilizers Inc., Haw River, NC, USA).

Physical characterization and analysis

The aerogel samples were weighed using a digital balance (U.S. Solid-Analytical Balance, Cleveland, OH 44103, USA). The balance was calibrated before each session to ensure accuracy. Measurements were repeated three times for each sample, and the averaged values with standard deviations were recorded to ensure precision.

To capture the intricate details of the samples, images were taken using a Galaxy Note 20 Ultra (Samsung, Pyeongtaek, Gyeonggi, South Korea). A scale was included in the field of view during imaging, allowing for subsequent calibration and dimensional analysis of the aerogels. The diameter and length of aerogels were measured using Fiji software by analyzing these microphotographs⁵⁰.

For a comprehensive assessment of the internal structure and morphology of the aerogels, micro-computed tomography (micro-CT) analysis was performed. The cylindrical aerogel samples were securely mounted in a specialized holder using a single-sided tape to ensure stability during scanning. A Bruker SkyScan 1276 – CMOS Edition micro-CT scanner (Kartuizersweg, Kontich, Belgium) was used, with scanning parameters meticulously configured to optimize resolution and imaging quality. Initial surface area and porosity analyses were conducted to examine the structural characteristics of the aerogel. Cross-sectional and longitudinal images were then captured through a rotation mechanism within the micro-CT scanner, with images taken from various angles to build a comprehensive representation of the internal structure. The micro-CT data was reconstructed using CTVox SkyScan software, which employed advanced algorithms to convert raw projection data into high-resolution 3D images. The visual analysis was further enhanced by generating 3D reconstructions and virtual slices, followed by in-depth analysis utilized CTVox software tools for quantitative metrices like voxel-based density distribution and porosity mapping, including open pores and close pores.

The cross-sectional morphology of the aerogel samples was further investigated using SEM (FEI Quanta 200, Hillsboro, OR, USA). The aerogel specimens were first attached to SEM stubs using a conductive adhesive for grounding, then sputter-coated with a thin layer of conductive material, such as gold, to augment surface conductivity and mitigate charging effects during imaging. This coating procedure was performed in an argon gas environment at a voltage of 110 milliamperes. The coated samples were placed in the SEM chamber under high vacuum conditions, with nitrogen gas present. Before imaging, the SEM parameters including acceleration voltage (25.0 kV), working distance, and aperture size, spot (3.0), and dwell time (1–3 µs)—were carefully adjusted to optimize imaging conditions for the aerogel’s cross-sectional morphology. SEM images were taken at various magnifications to capture the fine details of the internal structure, focusing on cross-sectional views to reveal the pore composition and distribution within the material.

Pore diameter measurements were conducted using ImageJ software, following the acquisition of clear, high-resolution SEM images of the aerogel samples. The images were calibrated using the scale bar where applicable, converting pixel measurements to real-world units. Image processing techniques, such as contrast and brightness adjustments, were applied to enhance pore visibility. Pore diameters were measured using Fiji’s ‘Straight Line’ tool, with a line drawn across a representative pore through its center. For calibrated images, the software provided measurements in real-world units; for non-calibrated images, known scale information was used to convert the measurements. Multiple measurements were taken across different pores and regions of the SEM image to ensure representativeness. The data analysis in Fiji generated a spreadsheet of the recorded measurements, from which the average pore diameter and standard deviation were calculated when multiple measurements were recorded. Likewise, the diameter and length of short NFs and MFs were quantified using Fiji analysis of their SEM images. The volume of the cylindrical aerogels or XStat® was determined using the Eq. (1).

In Eq. (1), “V” represents the volume of the cylindrical structures, “r” denotes the radius, and “h” signifies the height.

Mechanical properties

A compression test was conducted to assess the mechanical properties of the aerogel samples, which were cylindrical in shape. The testing was performed utilizing an Instron 5640 universal test machine (CellScale Biomaterials Testing, Waterloo, Ontario, Canada). The samples were secured within petri dishes using double-sided tape and then mounted onto the lower compression plate of a CellScale Univert apparatus (Serial Number: UV55290, CellScale Biomaterials Testing, Waterloo, Ontario, Canada). To simulate physiological conditions, the petri dishes were filled with citrated whole human blood (courtesy of the UNMC Blood Bank, Omaha, NE, USA). The compression test began by applying a load of 200N to induce compressive deformation at displacement levels of 20%, 40%, 60%, 80%, and 90%, all at a controlled rate of 1 mm/s. After each compression cycle, the samples were held for 5 min to allow for stress relaxation within the aerogel structure. To assess the shape recovery capabilities of the aerogel specimens, the compressive force was completely released after the 5-min holding period. The aerogels were then submerged in human blood following each compression levels of 20%, 40%, 60%, 80%, and 90%. The lengths of the aerogels were carefully measured before and after recovery using a digital caliper with a resolution of 0.01 mm. The study determined the maximum compressive strength at various deformation levels (20%, 40%, 60%, 80%, and 90%) after each compression cycle. To ensure consistency and reliability, three samples from the same batch were tested in each experiment. The specific elastic modulus was calculated using the following Eq. (2).

In Eq. (2), “E” represents the elastic modulus of the tested samples and “m” and “V” denote mass and volume, respectively. The volume of the cylindrical aerogels was determined using Eq. (1). The elastic modulus (E) was calculated using the following Eq. (3)⁵¹.

In Eq. (3), “F” represents the compressive force, “A” is the cross-sectional area of the aerogels, H₀ denotes the initial height, and ΔH indicates the change in length after compression.

We then aimed to investigate the impact of cyclic compressive strains on the mechanical properties, including strength and flexibility, of samples. The experiment involved subjecting each sample to cyclic compressive strains of 70%, 80%, and 90% over five rounds. Each round consisted of three repetitive cycles at the designated compressive strains, with each cycle involving 10 s of compression, 5 s of holding, 10 s of recovery, and 5 s of rest. The sample was first subjected to 70% compression, followed by 80% and finally 90% in subsequent rounds. To assess comparable metrics of compression resistance among the aerogels, we calculated critical forces, max forces, and force loss. Critical force, representing the force required to initiate compression, was determined as the maximum force observed at the peak of the force-displacement curve. Max force was defined as the maximum compressive force achieved during the specified displacement levels (70%, 80%, and 90%). Force loss was calculated as the percentage difference between max forces in consecutive cyclic compressions. Moreover, the change in length of the aerogels after cyclic compression tests was evaluated as the percentage difference between maximum changes in length across successive cycles. For comparison, different samples were subjected to cyclic compression. XStat®, NA, and MA samples underwent 5 cycles of cyclic compression at 70%, 80%, and 90% displacement, while NMA samples underwent 50 cycles of cyclic compression at 90% compressive strain to assess mechanical robustness. The compression, hold, recovery, and rest durations were standardized at 5, 1, 5, and 1 s per cycle, respectively, with a compression and relaxation rate of 1 mm/s.

To further investigate the influence of cross-linking time on the mechanical properties of NMA, cyclic compression-relaxation tests were conducted to assess their compressive strength and changes in length over multiple cycles. Two sets of NMA were prepared with different cross-linking durations: one set cross-linked for 6 h and the other for 18 h. Initially, both sets underwent cyclic compression tests for 10 cycles at 90% strain per cycle, with standardized duration for compression, hold, recovery, and rest at 5, 1, 5, and 1 s, respectively, and a compression and relaxation rate of 1 mm/s. Force measurements were recorded at 90% strain during compression. Subsequently, the aerogels with a 6-h cross-linking time were subjected to an extended test of 100 cycles to evaluate their mechanical robustness, maintaining the same compression protocol. The change in length after cyclic compression was assessed by measuring the percentage difference between the maximum changes in height observed across consecutive cycles.

To assess the toughness of XStat®, NA, MA, and NMA samples compression tests were conducted at a displacement level of 90%. Each test cycle included standardized durations for compression, hold, recovery, and rest, which were set at 10, 60, 10, and 60 s, respectively. The compression and relaxation rates were maintained at 1 mm/s throughout the test. Toughness was determined by analyzing the stress-strain curve and calculating the area under the curve using the Eq. (4).

Where “T” is the toughness, “σ” represents stress, “ϵ” denotes strain, and “ϵf” indicates strain at failure. We then conducted five cyclic compression tests to measure the hemostats’ elastic energy absorption and resilience. Each test involved compressing the samples to a strain of 90% at a speed of 1 mm/s. The compression cycle included four phases: compression, hold, relaxation, and rest, each lasting 10, 5, 10, and 5 s, respectively. To quantify the energy absorbed during compression at 90% strain, we recorded the maximum compressive force applied to the hemostats and the corresponding displacement. These measurements allowed us to calculate the stress and strain at various points during the compression process. The area under the stress-strain curves was then computed to determine the energy absorbed by the aerogels during compression.

Resilience (denoted as R) was calculated using Eq. (5) by determining the area under the stress-strain curve up to the elastic limit. This provides insight into the aerogel’s ability to store and release energy under loading and unloading.

Where “σy” and “ϵy” represent yield strength and yield strain.

Simulations

The mechanical properties of NA, MA, and NMA were evaluated using finite element method (FEM) simulations conducted with the Student Version of ANSYS 2024 R1. Compressive forces were applied along the Y-axis, and the resulting distributions of Von Mises stress were analyzed. To understand how the different architectures influence stress transmission, the applied force was normalized by the mass of each sample.

Additionally, FEM simulations for NMA and XStat® were conducted using fluid flow analysis in ANSYS Student 2024 R1. Blood considered an incompressible fluid in these simulations, was assigned a density of 1060 kg.m⁻³ with a zero rate of density change. The governing equations for these flow simulations included the continuity equation and the 3D incompressible Navier-Stokes equation, with the continuity equation expressed as Eq. (6).

Where p is density, v is velocity, and \(\nabla\) is gradient operator. The gradient operator was computed by the following Eq. (7).

In addition, the Navier-Stokes equation can be expressed as Eq. (8).

In Eq. (8), t is time, p is the fluid pressure, µ is the fluid dynamic viscosity, and f is the external forces applied to the fluid.

For the viscosity μ, the non-Newtonian behavior of blood flow is modeled using the Carreau model, which is given below as Eq. (9).

In Eq. (9), \({\mu }_{0}\) and \({\mu }_{\infty }\) are the zero and infinite shear rate viscosities, respectively, and λ is the relaxation time constant. For the case of blood, \({\mu }_{0}\) = 0.056 kg m⁻¹ s⁻¹, \({\mu }_{\infty }\) = 0.0035 kg m⁻¹ s⁻¹, \(\lambda\) = 3.313 s, n = 0.3568.

In all numerical simulation projects, a steady-state approach was used for the calculation. The blood flow velocity in the inferior was set at 10 m/s, with the flow assumed to be laminar. The outlet was designated as the outflow. The calculation parameters were as follows: a time step size of 0.01, 50 time steps, and a maximum 200 iterations per time step.

Fluids uptake behaviors

A series of experiments were conducted to assess the fluid absorption properties of NA, MA, and NMA in comparison to XStat®. Initially, the dry and compressed forms of each sample were exposed to either water or human blood (UNMC Blood Bank). The physical appearance of the samples immersed in blood was thoroughly documented using a digital camera (Apple 14 Pro Max, Cupertino, CA, USA), with the images presented unaltered. Additionally, the pH values of the diluted blood after absorption were measured using a pH meter (Orion Star^TM A221 Portable pH Meter, Thermo Scientific, Waltham, MA, USA).

For quantitative analysis, the compressed forms of NA, MA, NMA, or XStat® were weighed initially (denoted as “W_d” in g). These materials were then immersed in either water or human blood. After a precise interval of 30 s, the materials were removed and re-weighed (denoted as “W_W” in g). Given the variability in the dry weight of the aerogels (e.g., NA, MA, or NMA) and XStat®, the following Eq. (10) was employed to calculate the fluid absorption of the materials in terms of weight/weight.

To further evaluate the water and blood absorption capacities of the aerogels (e.g., NA, MA, and NMA) in comparison to XStat®, the study was extended to include volume variations. Initially, the volume of each material (NA, MA, NMA, or XStat®) was measured using Eq. (1), and recorded as V (cm³). The dry and compressed samples were then immersed in either blood or water. Their positions in the liquids were meticulously recorded using a digital camera (Apple 14 Pro Max). The samples were removed from the liquids at specific intervals —3, 6, 9, 12, 15, and 30 s for water, and 5, 10, 15, 20, and 25 s for blood. The water or blood absorption capacity was subsequently calculated using Eq. (11).

The water and blood absorption rate, expressed in g/cm³/s, were determined by calculating the slope of the water or blood absorption capacity versus time curve. This calculation was performed using Eq. (12).

Shape memory properties

A series of experiments were conducted to assess the shape-memory properties of NA, MA, and NMA compared to XStat® in both water and blood. The aerogels were initially sized at at 5.5 × 1 cm², while XStat® was sized at 3.3 × 1 cm². The aerogels underwent compression to form pellets. Subsequently, the NA, MA, and NMA pellets were exposed to either water or blood. The entire shape recovery process, along with the associated timeframes, was meticulously recorded using a digital camera (Apple 14 Pro Max). The volume of the samples before and after shape recovery was measured using Eq. (1). The shape recovery time, defined as the duration required for the aerogels to fully return to their original shape, was measured with a stopwatch (Fisherbrand^TM Traceable^TM Four-Channel Countdown Stopwatch with Memory Recall, Fisher Scientific). Given the variations in length between NA, MA, NMA (5.5 × 1 cm²) and XStat® (3.3 × 1 cm²), the shape memory time was normalized and calculated in s per cm using Eq. (13). This provided a standardized measure of the time each sample required to recover its shape per centimeter when immersed in water or blood.

A comparative analysis was performed to assess the shape recovery rate of dried NMA pellets in blood, comparing it with data on wet and dry materials reported in the literature over the past 20 years. This analysis involved a comprehensive search of PubMed, Scopus, and Google Scholar databases, focusing on materials known for their high mechanical resilience as hemostats for JH or NCTH. Keywords used in the search ‘shape memory materials’, ‘aerogels’, ‘hemostats’, ‘cryogels’, ‘sponges’, ‘foam’, ‘JH, ‘rapid hemorrhage’, ‘hemostasis’, and ‘NCTH’. The information gathered provided valuable insights into the shape recovery characteristics of NMA relative to established materials in the field.

In addition, experiments were conducted to evaluate the shape recovery percentages of NA, MA, and NMA in comparison to XStat® at various time points in blood. Initially, the aerogels were compressed to fix their shapes and then exposed to water and blood. The shape recovery percentages were recorded in water after 5 s of contact. For blood, samples were extracted at intervals of 5, 10, 15, 20, and 30 s to assess the shape recovery over time. The percentages of shape recovery in both water and blood were calculated using Eq. (14).

In Eq. (14), “L_R” is the recovered length at specific time points, and “L_T” is the total length of the sample. The shape recovery ratios in water and blood were documented using a digital camera (Apple 14 Pro Max).

The reversible absorption properties of water-soaked NMA were quantified using a manual compression method. The aerogels, with a diameter of 10 mm and a height of 7 mm, were compressed axially from the top. This compression process caused the aerogels to releaese water, which was then reabsorbed once the force was removed. The amount of water released, and the subsequent reabsorption were meticulously recorded using a digital camera (the Samsung Galaxy Note 20 Ultra 5G).

Additionally, the microstructure recovery of NA, MA, NMA, and XStat® before and after absorbing water was analyzed using SEM. This analysis also examined the relationship between shape recovery in human blood and the absorption of blood cells on the surface and cross-sectional areas at various time points. To prepare the samples for SEM analysis after blood absorption, the chemical-dry method was employed. Samples were first rinsed with 1× Dulbecco’s Phosphate Buffered Saline (DPBS) (Thermo Fisher Scientific, Waltham, MA, USA) and then fixed in a solution of 2% paraformaldehyde (PolySciences, Warrington, PA, USA) and 2.5% glutaraldehyde (Ladd Research, Williston, VT, USA) in 0.1 M Sorenson phosphate (Electron Microscope Sciences, Hatfield, PA, USA) for 3 h at room temperature. Following fixation, the samples were thoroughly washed with DPBS to remove excess fixative. The samples were then treated with 1% osmium tetroxide (Sigma-Aldrich) for 30 min at room temperature, followed by three additional washes with DPBS. For dehydration, a graded ethanol series (Fisher Scientific) was used, progressing through 35%, 50%, 70%, 95%, and 100% ethanol, with each concentration applied for 5 min. The samples were further treated with a graded series of hexamethyldisilazane (HMDS) (Electron Microscope Science) at 30%, 70%, and 100% concentrations, each for 5 min. The final step involved allowing the samples to air-dry in 100% HMDS within a chemical hood.

Pro-coagulant hemostatic efficacy evaluation

To test the hemostatic efficacy of the tested samples, a hemolysis assay was performed using a 2% (v/v) erythrocyte suspension. First, 2 mL of freshly anticoagulated whole human blood was diluted with 5 mL of normal saline and centrifuged at 100 × g for 15 min. The resulting erythrocytes were washed three times with normal saline, and then the pure erythrocytes were diluted to prepare a final 2% (v/v) erythrocyte suspension. Each sample was preheated in 0.8 mL of normal saline at 37 °C for 30 min. Next, 0.2 mL of the 2% (v/v) erythrocyte suspension was carefully added to the samples, followed by gentle mixing. The mixtures were incubated at 37 °C for 1 h, then centrifuged at 100 × g to separate intact erythrocytes. The supernatant was cautiously transferred to new tubes for photographic documentation, and its absorbance was measured at 540 nm using a microplate reader (Multiscan^TM FC Microplate Photometer, ThermoFisher Scientific, Waltham, MA, USA). Normal saline and deionized (DI) water served as the negative and positive controls, respectively. The hemolysis ratio was quantified using Eq. (15)⁵².

In Eq. (15), V_H, V_NC, and V_PC symbolize the absorbance of the supernatant corresponding to the samples (NA, MA, NMA, or XStat®), negative control, and positive control groups, respectively.

The hemostatic efficacy of NA, MA, and NMA was assessed by determining BCI, with QuikClot® Combat Gauze (QCG®) and XStat® used as comparative controls. For this evaluation, QCG® combat gauze and compressed shape-memory samples (e.g., NA, MA, NMA, and XStat®) were placed in EP tubes. Following a 10-min incubation at 37 °C, 50 μL of citrated whole human blood (UNMC Blood Bank) was carefully applied to the top surface of each sample. The samples were then incubated for an additional 10 min at 37 °C. Following this, 1.5 mL of DI water was added to each EP tube. The optical density of the resulting supernatant was measured at 540 nm (OD₅₄₀ nm) using a microplate reader (Multiscan^TM FC Microplate Photometer, ThermoFisher Scientific), and this measurement was recorded as the Value of Hemostats (VH_BCI). A control solution of mixed DIW and CWB (1.5 mL/50 μL) was used to establish a baseline reference. The OD₅₄₀ nm value from this negative control solution served as the Value of Control (VC_BCI), while VR_BCI defines the reference values for the subsequent calculations. The BCI was then computed using Eq. (16).

We next determined the whole blood clotting time for NA, MA, and NMA in comparison to QCG® and XStat®. In this experiment, 500 μL of whole blood containing 10% sodium citrate was added to a polypropylene tube containing each sample (NA, MA, NMA, XStat®, and QCG®). Then, 20 μL of a 0.25 mol L⁻¹ CaCl₂ solution was added to the tube. The tube was inverted every 5 s, and the onset of blood clotting was recorded at each interval.

The interactions between the samples and RBCs were also investigated, with QCG® and XStat® serving as commercial controls. Compressed samples of NA, MA, NMA, and XStat® and QCG® were placed in a 24-well microplate. Next, 100 μL of the RBCs suspension was applied to the top surface of each sample. After a 1-h incubation at 37 °C, the samples were rinsed with a phosphate buffer solution (PBS, pH = 7.4) to remove non-adherent RBCs. The samples were then transferred into 4 mL of DI water to lyse the adhered RBCs and release hemoglobin. After another 1-h incubation, 100 μL of the supernatant was extracted and placed into a 96-well microplate, where its optical density at 540 nm (VH_RBC) was measured. The OD₅₄₀ nm value of a solution containing 100 μL of RBCs suspension and 4 mL of DIW was used as the control value (VC_RBC), while VR_RBC defines as reference values. The percentage of adhered RBCs was calculated using Eq. (17).

A platelet adhesion assay was conducted to further investigate the interactions between various hemostats and platelets. To prepare for this assay, platelet-rich plasma (PRP) was obtained by centrifuging citrated whole blood at 100 × g for 15 min. Similar to the RBC assay, NA, MA, NMA, and XStat® were compressed, placed in a 24-well microplate, and 100 μL of PRP was applied to their top surfaces. For QCG®, 100 μL of PRP was applied on the top of it. After incubating the samples for 1-h at 37 °C, they were washed with PBS to remove non-adherent platelets. The samples were then soaked in a 1% Triton X-100 solution to lyse the adhered platelets and release the lactate dehydrogenase (LDH) enzyme. The optical density at 490 nm (OD₄₉₀ nm) of the resulting supernatant was measured (VH_Plt). The OD₄₉₀ nm value of a solution containing 100 μL of PRP that had not been exposed to the hemostats was used as the control value (VC_Plt), while VR_Plt was defined as reference values. The percentage of adhered platelets was calculated using Eq. (18).

To further investigate the adherence of RBCs and platelets on various hemostats, a detailed SEM (FEI Quanta 200, Hillsboro, OR, USA) analysis was conducted. In this analysis, samples of NA, MA, NMA, and XStat® were placed in individual wells of a 24-well microplate and exposed to 100 μL of RBCs and PRP suspensions. After a 1-h incubation at 37 °C, the samples were rinsed with 1×DPBS and fixed with a solution containing 2% paraformaldehyde and 2.5% glutaraldehyde (Ladd Research). The samples were thorough washed, treated with osmium tetroxide (Sigma-Aldrich), and then washed again. They were dehydrated using a graded ethanol series, and subsequently treated with a graded series of HMDS (Sigma-Aldrich). Finally, the samples were dried in 100% HMDS within a chemical hood.

To investigate the coagulation activation pathway of NA, MA, and NMA, as well as the commercially available products XStat® and QCG® (which served as positive controls), PT and aPTT assays were conducted. Citrated whole human blood was prepared by mixing whole blood with 3.8% sodium citrate in a 9:1 ratio, followed by centrifugation at 1500 × g for 15 min to obtain platelet-poor plasma (PPP). For the PT assay, each sample was incubated with 200 μL of PPP at 37 °C for 5 min. After incubation, 200 μL of PT reagent (Fisher Scientific) was added to the hemostats, and the time required for PPP to clot was recorded using a stopwatch.

Similarly, the aPTT assay involved incubating of each sample with 200 μL of aPTT reagent (Fisher Scientific) at 37 °C for 5 min. Following this, 100 μL of a 0.025 M CaCl₂ solution was added to activate the intrinsic coagulation pathway, and the clotting time taken was recorded using a stopwatch (Samsung Galaxy Note 20 Ultra).

Weighted ranking of aerogels

A weighted ranking methodology was used to select the best-performing aerogel samples for in vivo experiments⁵³. First, a Z-score transformation was applied to the raw data to standardize and normalize the datasets, making the results easier to interpret. The Z-score scale, with a mean of ‘0’ and a standard deviation of ‘1’, allowed for robust comparisons among NA, MA, and NMA. Each aerogel’s performance metrics were quantified and standardized using the Z-score formula, which involves subtracting the sample mean (µ) from the observed value (x) and then dividing by the sample standard deviation (σ), as delineated in Eq. (19).

Moreover, categorical and subjective weighting was applied to two critical criteria—physical characteristics and hemostatic properties—to refine the analysis. Physical characteristics included factors such as porosity, percentage of open pores, specific elastic modulus, shape memory, blood absorption rate, and the adhesion of RBCs and platelets. Hemostatic properties included variables like shape recovery time in human blood, BCI, blood clotting time, PT, aPTT, and hemolysis.

To provide a more precise evaluation of each aerogel’s overall performance, weightings were assigned to each parameter based on its relative importance. These weightings were determined through domain expertise and a detailed understanding of their significance. By combining these weighted factors with standardized Z-scores, each aerogel was ranked according to its performance, making it easier to identify the optimal designs. This approach facilitated direct comparisons among the various tested aerogels with different units, offering a clear framework for optimizing aerogel design for subsequent investigations. The weighted average was computed using Eq. (20),

Where x represents the weighted average, w_i signifies the sum of the product of the weight, and x_i denotes the data number.

Animal studies

The experimental design for the animal study adhered to the scientific rigor and transparency required by the National Institute of Health (NIH) and followed the ARRIVE guidelines (Supplementary Table 8)³². Approval was obtained from UNMC IACUC under protocol no. 22-051-08-EP, with all procedures conducted in strict compliance with UNMC IACUC guidelines.

The study involved inbred Yorkshire swine (3 months, both male and female, all non-castrated, n = 5), sourced from a UNMC-approved vendor. The animals were fasted for 12-h before surgery, with free access to water. On the day of surgery, premedication was administered, and an intravenous (IV) line was established in a marginal ear vein using a 20–22-gauge Angiocatheter (Nordson Medical, Loveland, CO 80538, USA). Anesthesia was induced via isoflurane and oxygen (3–5 L/min) using a veterinary anesthesia ventilator (MDS Matrx 3000, HALLOWELL EMC, Pittsfield, MA 01201, USA) to facilitate intubation. During the procedure, anesthesia was maintained via isoflurane and oxygen (1–2 L/min) until death or euthanasia. Vital signs were continuously monitored, with a rectal temperature probe and EKG (cardiac) monitors (DRE Waveline VS, M16C13140003, PET PRO SUPPLY CO.®, Frisco, TX, USA) in place. The swine were positioned on a water-circulated warming blanket (BLANKETROL® II, 222R, Cincinnati Sub-Zero Products, Inc., Cincinnati, OH 45241, USA), maintained at 103 ˚F. Mechanical ventilation was set at 12–15 breaths per min with a tidal volume of 5–10 mL/kg (MDS Matrx 3000, HALLOWELL EMC, Pittsfield, MA 01201, USA), and end-tidal pCO₂ was maintained between 35–45 mmHg. All incisions were made using an electrosurgical generator (Valleylab Force 1C, Pfizer, New Bedford, MA, USA).

For pressure monitoring and blood sampling, a 20-gauge Angiocatheter was surgically placed in the carotid artery and a 14–16-gauge Angiocatheter was inserted in the jugular vein for fluid and medication administration. These catheters (Nordson Medical, Loveland, CO 80538, USA) were inserted through a surgical cutdown in the right or left neck (Supplementary Fig. 18). Concomitantly, a midline laparotomy and splenectomy were performed to minimize autotransfusion from the contractile porcine spleen during stress⁴⁹, a mechanism that is not present in humans but crucial for a human-like hemorrhage model (Supplementary Figs. 19, 20 and Supplementary Movie 12). The spleen was weighted (Supplementary Fig. 21), and warm lactated Ringers (LR) solution was administered at three times the spleen’s weight at a rate of 100 mL/min using a roller infusion pump (DOSE IT P910, Integra Biosciences AG, Tardisstrasse, Zizers 7205, Switzerland). For a typical spleen weighing ~300 g, approximately ~900 mL of LR was given for post-splenectomy fluid replacement. A transabdominal cystostomy tube was placed through the bladder dome, secured with a purse-string suture, and exited through the lateral abdominal wall (Supplementary Fig. 22 and Supplementary Movie 13). This procedure controlled urine flow during the experiment and prevented undue pressure on the injury site, which could affect the hemorrhage rate. Afterward, the midline laparotomy incision was closed using towel clips, and the animals were covered with blankets to prevent hypothermia due to post-injury blood loss (Supplementary Movie 14).

To perform the surgical incision and identify the right femoral artery in domestic pigs, a midline incision was made in the right groin area using an electronic scalpel, exposing the underlying tissues. The right femoral artery and vein were chosen due to their diameter, which closely resembles that of human vessels, ensuring that the bleeding force in the swine closely mimicked human conditions. Subsequently, the subcutaneous tissue in the pig was carefully dissected until the femoral vessels were visually identified (Supplementary Fig. 23 and Supplementary Movie 15). The right femoral artery and vein were then clearly identified, and a deliberate injury to these blood vessels was induced by surgically transecting the femoral artery and vein under general anesthesia. This procedure allowed for a controlled 30-s period of free bleeding, simulating the conditions of a junctional injury similar to what might be encountered in a warfighter, such as a gunshot wound to the groin. The experimental design included four groups, each consisting of five pigs (the sample size was chosen according to a power analysis), to systematically investigate the impact of different treatments on the induced injury:

1. 1.

   Injury Only (Non-Treated Control): This group serves as the non-treated control, where no specific treatment is applied to the induced injury. The purpose is to showcase the inherent lethality of an uncontrolled injury, providing a baseline for comparison with treated groups.

2. 2.

   Injury with XStat® Treatment: Inclusion of this group involves administering the current primary comparator treatment for JH, utilizing XStat® pellets, in accordance with Tactical Combat Casualty Care (TCCC) guidelines⁷. This treatment is established as the standard for comparison due to its recognized effectiveness.

3. 3.

   Injury with QCG® Treatment: This group functions as a secondary comparator treatment, employing QCG®—a widely available standard issue in the U.S. Army. It represents an alternative intervention for JH, especially when XStat® is unavailable, providing a practical comparison to the primary treatment.

4. 4.

   Injury with NMA Treatment: This group constitutes the primary experimental treatment, involving the application of dried bicomponent aerogel pellets. The purpose is to explore the efficacy of the shape memorable NMA as an experimental intervention for JH. This treatment offers a unique approach, distinct from wet or biologically augmented treatments in the published reports, allowing for a comprehensive evaluation of its potential effectiveness.

Two devices of each type such as QCG®, XStat®, or NMA were applied to the injury site. For NMA, each device contains 35–50 compressed pellets, whereas each device of XStat® contains ~100 pellets. The application times of each application material were recorded using a stopwatch and the effects of these treatment groups were meticulously evaluated in terms of primary and secondary endpoints during a 180-min observation period conducted under general anesthesia. The selection of the 180-min time point was based on the time required for transferring patients to hospitals. Primary endpoints included the assessment of post-treatment blood loss, survival status at 180 min post-injury, and the incidence of post-treatment rebleeding.

The blood collected during or after compression, including any bleeding during the compression period, was directed into a separate suction canister. This separation facilitated the differentiation of blood loss before and after treatment. Following treatment, a new set of surgical gauze was employed for compression over a 3-min, during which the gauze absorbed blood. The manual pressure was applied with adequate force to stop the bleeding. Specifically, for this groin injury model, pressure was applied using both hands: the right hand applied direct and deep pressure on the hemostatic material, while the left hand pushed on top of the right hand to reinforce the force. Although there is no standardized or published methodology to precisely calibrate hand pressure in a model such as this, the combined force was likely in the range of 15–20 lb based on the estimation of our surgeons. Additionally, the application materials played a role in blood clotting by absorbing the whole blood within the wound. Considering the collective blood absorption by both the application materials and the surgical gauzes during or after compression as part of the post-treatment blood loss, the following Eq. (21) was employed for accurate quantification.

Where WC_b’ represents the weight of the canister used to collect blood lost after treatment until death or after 180 min of observation, and WC_e denotes the empty weight of the canister. WG_w‘ is the weight of wet gauze used for post-treatment manual compression for 3 min, and WG_d is the dry weight of the gauze. WM_b represents the weight of application materials after treatment, and WM_d is the weight of the materials before treatment.

The hemostasis percentages for each group depict the proportion of animals that ceased bleeding within a specified time frame, as determined by Eq. (22).

Here, Eq. (22) defines N_H as the number of swine stopped bleeding after treatment, while “n” represents the total number of experimental animals in the group. The hemostatic time for each application material was measured using a stopwatch, representing the total time each material took to stop bleeding after treatment. Rebleeding time, defined as the cumulative time of all rebleeding incidences for each subject after treatment, was recorded using a stopwatch. The percentage incidence of rebleeding in each group was calculated using Eq. (23).

In Eq. (23), N_rb represents the total number of experimental animals experiencing post-treatment rebleeding, while ‘n’ represents the total number of experimental animals in the group.

The hemostatic time and rebleeding percentages following treatment with NMA were systematically compared with relevant literature from 2010–2024. Our search encompassed databases such as Google Scholar, PubMed, and Scopus, focusing on studies on swine or pig artery and/or vein transection hemorrhage models. The search terms included swine, pig, hemostatic agent or materials, and hemostasis in swine artery and/or vein transection models. The findings from these searches were compiled and summarized in Supplementary Table 9, detailing the referenced papers’ experimental designs, models, and conditions. This table elucidates the comparability of these reference papers with the present study. Specifically, it highlights instances where the models and experimental conditions of the reference papers align with or are even less aggressive than those employed in this study.

Secondary endpoints included the assessment of final vital signs (i.e., mean arterial pressure and heart rate), hematologic parameters (e.g., hemoglobin, hematocrit, platelet count, and lactate levels), coagulation profiles (e.g., prothrombin time, partial thromboplastin time, international normalized ratio [INR], and fibrinogen levels), and various arterial blood gas parameters (e.g., bicarbonate [HCO₃], partial pressure of oxygen [PaO₂], partial pressure of carbon dioxide [PaCO₂], and end-tidal carbon dioxide [EtCO₂]), as well as pH and temperatures.

Mean arterial pressure, heart rate, and body temperatures were monitored by a designated technician from UNMC Comparative Medicine, who was blinded to both samples and subjects at predetermined time intervals during the experiments (e.g., 0 min, 15 min, 30 min, 60 min, 120 min, and 180 min) using an in-house EKG monitor (DRE Waveline VS, M16C13140003, PET PRO SUPPLY CO., Frisco, TX, USA).

Hematologic, coagulation, and arterial blood gas analyses were conducted at UNMC Hospital Pathology by pathologists unaware of the study details. Complete blood counts were performed using a Sysmex XN 9100 instrument (Sysmex America, Mundelein, IL, USA) at specified time points during the experiment (0 min, 15 min, 30 min, 60 min, 120 min, and 180 min). Lactate levels were measured using a Beckman Coulter AU5800 (Beckman Coulter Inc., Higashino, Shizuoka, Japan) machine at the same time intervals. Arterial blood gas values were assessed using a Radiometer ABL800 FLEX Analyzer (Radiometer Medical ApS, Aakandevej, Bronshoj, Denmark) at UNMC Hospital Pathology to measure HCO₃, PaO₂, PaCO₂, and EtCO₂ at various time points during the experiments. Coagulation parameters, including prothrombin time, partial thromboplastin time, INR, and fibrinogen levels, were determined using ACL TOP 500 instruments (Instrumentation Laboratory Company, Bedford, MA, USA) at different time intervals throughout the study. As mentioned above, the sample size was n = 5 for the Control, NMA, XStat®, and QCG® groups. However, for the baseline measurements of lactate level, the sample size was n = 4 for the Control, QCG®, and NMA groups due to insufficient sample quantity, instrumental error, or clotting of the blood prior to measurement (Supplementary Fig. 16). Similarly, the sample size was n = 4 for the Control group (for the baseline measurements of PT, aPTT, INR, and fibrinogen) and QCG® (for the baseline measurements of PT, aPTT, and INR) due to the same limitations (Supplementary Fig. 17).

In instances where a subject succumbed to hemorrhage before the 180-min mark, final secondary endpoints were obtained during the pre-terminal phase. Following the designated observation period, each pig underwent euthanasia through the intentional transaction of a large blood vessel in the chest, facilitating terminal hemorrhage (Supplementary Movie 16).

Data adoption for comparison

The data for the control, XStat®, and QCG® groups were adopted from our recent report to compare with the NMA treatment group¹. This method was selected to adhere to the 3Rs principle of animal research by minimizing the number of animals used while maintaining robust control data, as the data were collected under the same experimental protocol^54,55.

Power analysis

The sample size for the in vivo study was determined using power analysis, aimed at assessing survivability in a non-survival JH swine model. We hypothesized that without treatment, 90% of subjects would perish, whereas with NMA treatment, 90% would survive, with a 10% mortality rate. Based on an expected mortality rate of 10% in Group 1 (no treatment, control) and 90% in Group 2 (NMA treatment), with an alpha level of 0.05 and a beta level of 0.2 (power of 0.8), the necessary sample size was calculated. The sample size was determined using a two-sided test for comparing two proportions, specifically to detect a clinically significant difference in survival rates between the untreated and NMA-treated groups⁵⁶. The calculation was based on the following equation:

In this Eq. (24), the proportion (incidence) of subjects in groups #1 and #2, denoted as p₁ and p₂, respectively. Additionally, the absolute difference between these proportions, Δ, was calculated to quantify the expected disparity in outcomes between the two groups. The probabilities of type I and type II errors, denoted as α and β, respectively, were set at standard levels of 0.05 and 0.2. Critical Z values corresponding to these error probabilities were utilized in the sample size calculation. Moreover, the ratio of sample size between group 2 and group 1, denoted as “k”, was considered to ensure balance and efficiency in the study design. The values of q₁, q₂, \(\bar{p}\), \(\bar{q}\), and ∆ were calculated using the respective equations.

Based on these parameters, the analysis determined that a sample size of 5 animals per group was required to achieve an alpha level of 0.05 and a beta level of 0.2 (power of 0.8). The in vivo study included four groups: Control (no treatment), XStat®, QCG®, and NMA.

Statistical analysis

The Shapiro–Wilk test was used to assess data normality, and the data were presented as mean values with corresponding standard deviations (s.d.). Statistical analyses were conducted utilizing analysis of variance (ANOVA) with GraphPad Prism (version 9.5.1). Pairwise comparisons were carried out using either ordinary one-way or two-way ANOVAs, followed by Tukey’s multiple comparisons post hoc test when appropriate. Additional statistical comparisons were performed as indicated. For measurements obtained from photographs or SEM images, Image J was utilized after calibrating pixels to millimeters or micrometers. Survival analysis was performed using the log-rank tests to compare survival curves among groups. The Fisher exact test was employed for categorical data comparisons to assess statistical significance. All analyses were conducted with a significance level set at 0.05. Graphs and figures were created and visualized using GraphPad Prism, Origin Pro (version 8.5), and BioRender.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.