Plasma-deposited tetraglyme surfaces greatly reduce total blood protein adsorption, contact activation, platelet adhesion, platelet procoagulant activity, and in vitro thrombus deposition Lan Cao,1 Mark Chang,2 Chi-Ying Lee,1 David G. Castner,1,2 Sivaprasad Sukavaneshvar,3 Buddy D. Ratner,1,2 Thomas A. Horbett1,2 1 Department of Chemical Engineering, University of Washington, Seattle, Washington 98195 2 Department of Bioengineering, University of Washington, Seattle, Washington 98195 3 Utah Artificial Heart Institute, Salt Lake City, Utah 84103 Received 30 May 2006; revised 15 August 2006; accepted 29 August 2006 Published online 18 January 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.31091 Abstract: The ability of tetraethylene glycol dimethyl ether (tetraglyme) plasma deposited coatings exhibiting ultralow fibrinogen adsorption to reduce blood activation was studied with six in vitro methods, namely fibrinogen and von Willebrand’s factor adsorption, total protein adsorption, clotting time in recalcified plasma, platelet adhesion and procoagulant activity, and whole blood thrombosis in a disturbed flow catheter model. Surface plasmon resonance results showed that tetraglyme surfaces strongly resisted the adsorption of all proteins from human plasma. The clotting time in the presence of tetraglyme surfaces was lengthened compared with controls, indicating a lower activation of the intrinsic coagulation cascade. Platelet adhesion and thrombin generation by adherent platelets were greatly reduced on tet- raglyme-coated materials, compared with uncoated and Biospan1-coated glass slides. In the in vitro disturbed blood flow model, tetraglyme plasma coated catheters had 50% less thrombus than did the uncoated catheters. Tetraglymecoated materials thus had greatly reduced blood interactions as measured with all six methods. The improved blood compatibility of plasma-deposited tetraglyme is thus not only due to their reduced platelet adhesion and activation, but also to a generalized reduction in blood interactions. Ó 2007 Wiley Periodicals, Inc. J Biomed Mater Res 81A: 827–837, 2007 INTRODUCTION proteins are expected to have improved hemocompatibility. Human blood contains a variety of protein types, and the concentration of each type of protein varies greatly. For example, albumin, IgG, and fibrinogen (Fg) are the most abundant proteins in plasma, with a concentration of about 40, 10, and 3 mg/mL, respectively, whereas the concentration of von Willebrand’s factor (vWf) is only about 0.01 mg/mL. Blood plasma proteins also have different adsorption affinities that vary with the surface chemistry of materials. To distinguish which proteins are the major players in mediating platelet-induced clotting events, our group has studied separately the role of proteins in mediating platelet adhesion. We found that Fg and vWf are the two major proteins mediating platelet attachment, depending on flow conditions.11,12 The amount of adsorbed Fg needed to support full-scale platelet adhesion under both static and shear conditions11,12 was less than 10 ng/cm2, while even a very low adsorption of vWf (2 ng/cm2) was sufficient to greatly increase platelet adhesion at higher shear rates.12 To reduce the reactivity of artificial materials with human blood, many investigators have focused on preparing surfaces that would resist blood proteins adsorption.1–8 The rationale for this approach is that adsorbed proteins can initiate the intrinsic coagulation cascade, and more importantly, mediate platelet attachment and activation, either of which will ultimately lead to thrombotic and thromboembolic complications.9,10 Therefore, surfaces that can resist the adsorption of blood plasma Correspondence to: T.A. Horbett; e-mail: horbett@cheme. washington.edu Contract grant sponsor: NIH; contract grant numbers: HLR01, HL 67923 Contract grant sponsor: NIBIB; contract grant number: EB002027 ' 2007 Wiley Periodicals, Inc. Key words: tetraglyme; protein adsorption; platelet adhesion; platelet activation; plasma deposition 828 Previous work by López et al.13,14 and the current authors15,16 showed that a surface coating made by radio frequency plasma deposition of tetraethylene glycol dimethyl ether (tetraglyme) was able to greatly reduce Fg adsorption, down to an ultralow level (<5 ng/cm2 from plasma). However, it was not clear whether the high resistance to Fg adsorption was due to the preferential adsorption of other plasma adhesive proteins such as fibronectin (Fn) and vitronectin (Vn), that might outcompete Fg adsorption at the surface, or due to a general reduction of adsorption of all proteins.17 Note that at the levels Fn and Vn usually adsorb from plasma they did not contribute to platelet adhesion.18 However, it was still possible that the adsorption of these or other proteins in plasma could be elevated on the tetraglyme plasma coatings due to the lower Fg adsorption, and thus might play an increased role in platelet attachment. Bremmell et al.19 found Fg adsorption was substantially reduced on similar plasma-deposited poly (ethylene oxide) (PEO)-like coatings from single component Fg solutions, which cannot be explained by increased adsorption of competing proteins, thus suggesting that a general reduction of adsorption might also be responsible for lowered Fg adsorption to tetraglyme coatings. Nonetheless, we wanted to confirm that the adsorption of the other proteins in plasma was reduced on tetraglyme coatings. Therefore, a study of total blood protein adsorption on the tetraglyme plasma coating was done with surface plasmon resonance (SPR). SPR has been extensively used to study the dynamics and kinetics of protein adsorption.20,21 It offers the advantages of in situ, real time monitoring and avoids the need for a large series of studies with numerous radiolabeled proteins. SPR detects adsorption through the changes in refractive index at the surfaces, and since most plasma proteins have similar refractive indices, SPR measurements can be used to measure the total protein adsorption. The ability of ultralow Fg adsorption tetraglyme plasma-deposited coatings to reduce interactions with blood was also studied with five additional in vitro methods, namely clotting time in recalcified plasma, platelet adhesion and procoagulant activity, and whole blood clotting in a disturbed flow catheter model. Although prior studies showed that platelet adhesion to tetraglyme-coated surfaces was lowered, it was possible that tetraglyme coatings would activate the intrinsic coagulation cascade, and so a clotting time study was used to test this possibility. Similarly, if the low platelet adhesion to tetraglyme was accompanied by a higher activation of adherent platelets, tetraglyme coatings might be less bloodcompatible than expected; therefore, we assessed the procoagulant activity of adherent platelets. Finally, in vitro thrombus deposition was measured as an end-point indicator of the hemocompatibility of tetraglyme coatings. Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a CAO ET AL. MATERIALS AND METHODS Samples Microscope cover glass slips (10-mm diameter, Fisherbrand1, Cat. no. 10 CIR-1, Fisher Scientific, Pittsburgh, PA) were used as the substrate for Biospan1 samples. The glass disks were cleaned by sonication for 15 min in Isopanasol (C. R. Callen, Seattle, WA) solution with a ratio of 1:64 in DI water, and then air-dried in a laminar flow hood and stored under nitrogen until use. Biospan (Polymer Technology Group, Berkeley, CA) was dissolved in N,N-dimethylacetamide (Aldrich, Milwaukee, WI) and made into 1% (wt) solutions. The solutions were then filtered by syringe-driven filter units (Millex1 Millipore, Billerica, MA). The glass disks were coated using a commercial spin coating apparatus (Headway Research, Garland, TX). About 20 mL Biospan solution was put onto one side of glass disks, and they were then spun at 4000 rpm for 30 s. After drying in the laminar flow hood for at least 45 min, the other side of glass discs was coated by the same procedure. Polyethylene (PE) samples are 0.5-mm-thick low-density PE films (Laird plastics, Seattle, WA). IntramedicTM Clay AdamsTM PE tubing (1.14 mm ID, 1.57 mm OD) was from Becton Dickinson (Sparks, MD) as catheter samples. SPR chips used were SF14 glass slides (Schott Glass Technology, Durea, PA) coated with 2-nm Cr and 50-nm Au (99.99%), and then coated with tetraglyme plasma coating. Plasma deposition Tetraglyme was purchased from Sigma-Aldrich (Milwaukee, WI). Plasma deposition was conducted using the system previously described.22 Flat-disk samples were placed onto a glass substrate and coated on both sides in two deposition runs. For the SPR substrates, a 100-nm-thick film was deposited onto the gold-coated surface. Catheter samples were hung on a glass rack, so that the outside surfaces of catheters were fully exposed to the glow. Prior to deposition, argon etching at a pressure of 350 mTorr, flow rate of 4.0 sccm (standard cubic centimeters per minute), and power of 40 W was used to clean the substrate and generate free radicals for subsequent plasmapolymerized grafting to the substrate. Tetraglyme vapor was then introduced and maintained at an optimum flow rate (1.33 sccm) and pressure (350 mTorr). Power used during the tetraglyme deposition was 80 W for the first 1 min, 40 W for the next 30 s, and 10 W for the final 10 min. Samples were rinsed with DI water and blown dry with argon gas before and after deposition. Rinsing was shown not to cause delamination of the coating or to alter the chemical structure of coating. Tetraglyme plasma coated samples were analyzed by electron spectroscopy for chemical analysis (ESCA) as described previously,16 and all the test samples exhibited a C 1s ether (286.5 eV)-hydrocarbon (285.0 eV) ratio greater than 60%. Buffers and biochemicals Protein adsorption buffers included citrate phosphatebuffered saline (CPBS), which contained 10 mM Na2HPO4, 10 mM citric acid, and 120 mM NaCl. CPBS with azide IMPROVED HEMOCOMPATIBILITY OF TETRAGLYME PLASMA COATINGS (CPBSz) was CPBS with 0.02 % w/v added sodium azide. CPBSzI was CPBSz with added 10 mM NaI. PBS contained 10 mM Na2HPO4 and 120 mM NaCl. Platelet suspension buffer without metal ions (PSB-MIF) contained 5.5 mM dextrose, 4 mg/mL bovine serum albumin (BSA), 137 mM NaCl, 2.7 mM KCl, 0.4 mM sodium phosphate monobasic and 10 mM HEPES, and was sterile-filtered and supplemented with 0.1 U/mL apyrase, grade V from potato (Sigma-Aldrich, St. Louis, MO) prior to use. Platelet suspension buffer with metal ions (PSB-MI) was PSB-MIF with added 2.5 mM CaCl2 and 1.0 mM MgCl2. All the buffers were at pH 7.4. The lactate dehydrogenase (LDH) assay was done with a solution containing a proprietary mixture of lactate, NADH, iodotetrazolium chloride, and catalyst from Boehringer Mannheim (Indianapolis, IN). Human thrombin, factor Xa, factor Va, and prothrombin were from Enzyme Research Labs (South Bend, IN). S-2238 was a chromogenic substrate specifically for thrombin from Diapharma Group (Franklin, OH). The chemical structure of S2238 is H-D-phenylalanyl-L-pipecolyl-L-arginine-p-nitroaniline. Surface plasmon resonance The SPR liquid sensing system used in this study has been described and characterized in detail elsewhere.23 Briefly, the SPR system was set up with a planar prism (Kretschmann) configuration. A 50-nm gold layer on top of a 2-nm chromium adhesion layer on the glass slide was electron-beam evaporated at a pressure below 1 10–6 Torr. The glass side of the gold-coated substrate was refractive index-matched to the prism, while the tetraglyme-coated gold surface was mechanically pressed against a milled Teflon flow cell. A polychromatic light beam was passed through the prism and the backside of the goldcoated substrate, to excite surface plasmon waves at the metal-dielectric interface. The wavelength at which resonant excitation occurred depended on the refractive index of the analyte in the proximity of the SPR surface. Thus, the amount of adsorbed analyte at the surface can be quantified by measuring the shift in the resonant wavelength induced by the refractive index change. The reflected light was analyzed with a spectrograph, which determined the intensity versus wavelength at a fixed angle. During SPR measurements, the gold substrates were first equilibrated with degassed water followed by PBS buffer solution. After establishing a stable baseline, blood plasma was delivered to the flow cell at a flow rate of 50 mL/min at room temperature. Protein adsorption at the surface was observed by monitoring the wavelength shift of the SPR reflected minimum. The index of refraction of the various bulk solutions was measured with a spectronic Abbe-3L refractometer (Thermo Electron Corporation, Waltham, MA). Protein radioiodination and adsorption Purified human fibrinogen (Fg) and human von Willebrand’s factor (vWf) were purchased from Enzyme Research Laboratories. Human plasma was obtained from George King Bio-Medical (Overland Park, KS) and kept at –808C until use. BSA fraction V (Sigma-Aldrich, St. Louis, MO) was used to block nonspecific platelet adhesion. Radiolabeling of Fg with 125I was done with a modified ICl method using a 2:1 molar ratio of ICl to Fg, as previously 829 described.11 Radiolabeling of vWf with 125I was performed using iodo-beads (Pierce Biotechnology, Rockford, IL) as described.24 D-Salt desalting columns (Pierce Biotechnology) were used to remove excess unincorporated 125I from the iodinated proteins. Radiolabeled Fg and vWf were stored at – 808C and used within 2 weeks. The effect of radioactive decay was taken into consideration for all calculations. Protein solutions with specific activity of at least 100 cpm/ ng were prepared by adding 125I protein to 1% human plasma. The amount of added 125I protein was enough to provide the desired specific activity. The added 125I protein did not significantly change the concentration of the protein in the plasma, because of the high specific activity of the radiolabeled protein stock, that is, the mass of added 125I protein was less than 1% of the amount of the protein in the plasma. Before adsorption, samples were incubated with CPBSzI buffer for 45 min. Then protein solution containing 125I protein was added to the buffer solution in which the samples were still submerged, thus avoiding exposure of the samples to the air–protein solution interface. To adsorb proteins to tubular samples prepared for the catheter studies, the ends of the samples were sealed during adsorption, and protein was not allowed to adsorb to the inside surface of the tubes. Protein adsorption lasted for 1.5 h at 378C. Unbound protein was rinsed away by repetitively dipping sample disks or catheters into a beaker containing fresh CPBSzI buffer. The rinsing buffer was changed several times, until no radioactivity was detected in the rinse buffer. The radioactivity of surface-adsorbed proteins was then measured using a Cobra1 II Series Auto-Gamma1 Counting system (Packard Instrument, Meriden, CT). The amount of protein adsorbed onto the surface was calculated from the retained radioactivity, corrected for background, divided by the specific activity of the protein solution and the planar surface area of the sample. Platelet adhesion Platelets were collected by sequential centrifugation (180g for 20 min to collect platelet rich plasma (PRP), followed by 1500g for 15 min) of acid citrate dextrose (ACD) anticoagulated (1:9 v/v) human whole blood. The platelet pellets from the second centrifugation were resuspended in PSB-MIF. Platelet concentration was determined with a Cell-Dyn1 3700 cell counter (Abbott Laboratories, Abbott Park, IL) at the University of Washington Medical Center. Platelet concentration for adhesion studies was adjusted to be 1 108/mL in PSB. Metal ions (Ca2þ and Mg2þ) were added to the platelet suspension immediately prior to platelet adhesion experiments, to achieve a final concentration of 2.5 mM CaCl2 and 1.0 mM MgCl2. Flat samples were put into 24-well plates and preadsorbed with 1% human plasma for 2 h at 378C and blocked with BSA. The use of BSA for ‘‘blocking’’ on tetraglyme coating samples was to make the platelet adhesion condition comparable for the noncoated and tetraglyme-coated samples. After preadsorption with 1% human plasma, there should be negligible amount of BSA adsorbed to tetraglyme coatings (as suggested by SPR study), and so the BSA blocking will not affect the platelet adhesion results. After rinsing to remove unbound proteins, a platelet suspension was introduced and incubated at 378C for 1.5 h. After the incubation, unbound platelets were rinsed away with PBS buffer. The adherent platelets were lysed by adding Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a 830 CAO ET AL. 50 mL of 1% Triton in PSB to each well. The number of adherent platelets was determined by LDH assay of the Triton lysate, converting the observed LDH activity to the number of platelets with a calibration curve. Scanning electron microscopy (SEM) was also used to observe the morphology of adherent platelets. Thrombin generation Thrombin generation was measured using the previously published prothrombinase assay.25,26 Samples preadsorbed with 1% plasma and blocked with BSA were incubated with a platelet suspension for 1.5 h to allow platelets to adhere, using the same protocol described earlier in the adhesion experiments. After rinsing away the unbound platelets, the sample disks with adherent platelets were incubated with a mixture of human factor Va (6 nM) and Xa (150 pM) in PSB for 15 min, to allow prothrombinase to form on the platelet surface. Prothrombin (600 nM) was then added, and the solution was mixed well. After 0, 3, and 6 min, an aliquot (25 mL) was collected and put into separate wells of a 24-well plate containing EDTA (50 mL, 80 mM). The EDTA chelates Ca2þ and, therefore, stopped the conversion of prothrombin to thrombin. To assay the amount of thrombin formed, triplicate portions of the solution was transferred from each well of the 24-well plates to separate wells of a 96-well plate and then s-2238 (420 mM) was added to each well. The absorbance at 405 nm was recorded over a 5-min period, and mOD405 nm/min was calculated by the microtiter plate reader. Thrombin concentration was determined by dividing the mOD405 nm/min by the slope of a calibration curve made by plotting mOD405 nm/ min versus thrombin for a series of known thrombin solutions. Finally, the thrombin generation rate by the samples was calculated from the thrombin concentrations measured after 0, 3, or 6 min of incubation with the prothrombin solution. Recalcified plasma clotting time The clotting time protocol was similar to that reported by Grunkemeier et al.27 Human platelet poor plasma (citrate-anticoagulated) was brought to 20 mM CaCl2 by addition of calcium from a 1M stock solution. The plasma was then quickly mixed well by vortex, and 0.25 mL plasma was immediately added to the wells of a 24-well plate. Falcon 24-well nontissue culture treated polystyrene (PS) plates were used. Prior to adding plasma, the different surfaces to be tested had been placed into the wells. The surfaces used were 10-mm-diameter circular glass coverslips and fluorinated ethylene propylene (FEP) disks, either unmodified or tetraglyme-treated. Twentyfour-well plates were placed in a 378C shaking water bath, and the clotting time of the plasma was determined visually. The plasma clotting time was measured as the time it took for the plasma to undergo gelation, detected by loss of movement of the plasma in response to the rotation and shaking. Blood interaction studies in an in vitro catheterization model An in vitro model that incorporated disturbed blood flow was utilized to assess the efficacy of tetraglyme-coated catheJournal of Biomedical Materials Research Part A DOI 10.1002/jbm.a Figure 1. Schematic illustration of the in vitro model to measure thrombus deposition on catheters. Catheters were either untreated or tetraglyme plasma treated 1-mm internal diameter PE tubes, sealed at the ends. The tetraglyme coating was on the outside surface of the PE tubes. ters in reducing catheter thrombosis.28 The samples evaluated were plain 1.57 mm OD PE tubes, or PE tubes coated on the outside surface with tetraglyme. Fresh bovine blood was obtained by inserting a cannula directly into the heart of stunned cows from a local abattoir. Approximately 4–10 L of blood was collected into a collapsible reservoir and anticoagulated with heparin (1.5 U/mL final concentration). Autologous 111indium-labeled platelets were added to the blood, and the blood from a single animal was divided into two separate 1-L blood reservoirs. To make the radiolabeled platelets, 200 mL of autologous blood was collected prior to the commencement of experiment in ACD and centrifuged at 350g for 15 min to sediment the red cells. The supernatant platelet-rich plasma was separated and centrifuged at 1000g for 15 min to sediment platelets. After decanting the plasma, the sedimented platelets were resuspended in 5 mL ACD-saline (0.25 mL ACD þ 4.75 mL of 0.9% NaCl) to which 50–200 mCi 111indium oxine (Intermountain Radiopharmacy, Salt Lake City, UT) was added and incubated at 378C for 20 min. As shown in Figure 1, uncoated and tetraglyme plasma coated 15 cm 1-mm ID PE tubing was deployed one at a time inside 3.2 mm ID PVC Tygon1 tubing (Saint-Gobain Performance Plastics, Akron, OH) by inserting them through the wall of the PVC tubing, to simulate catheter deployment in a blood vessel. The tubing segments with the devices were connected IMPROVED HEMOCOMPATIBILITY OF TETRAGLYME PLASMA COATINGS Figure 2. SPR measurements of total protein adsorption from various concentrations of human blood plasma to uncoated and tetraglyme plasma coated SPR gold chips. For all curves, the solution was changed from buffer to plasma and then to buffer again. (a) 1% plasma on bare gold; (b) 1% plasma on tetraglyme; (c) 10% plasma on tetraglyme; (d) 50% plasma on tetraglyme; (e) 100% plasma on tetraglyme. Curves for tetraglyme have been corrected for the effect of the refractive index change caused by the overlayer of tetraglyme film (see text). to the blood reservoirs (maintained at 378C). A roller pump was used to induce blood flow at 75 mL/min for 1 h. At the end of the experiment, the test devices were retrieved, and radioactivity associated with the thrombi on the device was measured with a gamma counter (Minaxi 5000, Packard). Statistical analysis 831 that the proteins adsorbed on bare gold was strongly bound. When the tetraglyme sample was exposed to higher plasma concentrations, there was an increased wavelength shift (18 nm for 10% plasma, 22 nm for 50% plasma, and 71 nm for 100% plasma) in the presence of plasma. After the buffer rinse, the wavelength shift was greatly reduced. Even after exposure to 100% plasma, the wavelength shift after the rinse step was <1 nm (equivalent to a total mass of adsorbed proteins of 24.1 ng/cm2). This indicated that even at a high protein concentration, the tetraglyme coatings were still able to keep most proteins from adsorbing. The greater wavelength shift in the presence of plasma was probably due to the coupling of the evanescent wave with the high concentration of bulk phase protein. It should be noted that the layer of tetraglyme plasma coatings changed the sensitivity of measurements of wavelength shifts, because the attenuation of the evanescent wave by protein adsorption to tetraglyme is less than for protein adsorbed to gold. Thus, the refractive index change caused by protein adsorption on the tetraglyme-coated gold surface will be less than on a bare gold surface. Using the equation by Jung et al.,23 we estimated that, on a typical 100-nm-thick tetraglyme film (measured by AFM, images not shown here), the shift caused by an adsorbed protein film would be 35% less than that by the same protein on gold. All wavelength shifts on the tetraglyme plasma coated sample presented here were therefore corrected to account for this attenuation. The total mass of adsorbed proteins can be approximately estimated using the following two equations developed by Jung et al.:23 Coated gold: R ¼ mðZa Zs Þ½1 expð2da =ld Þ Comparison between two groups were made by the Student’s t test. Values of p < 0.05 were considered significant. RESULTS Surface plasma resonance The SPR studies showed that tetraglyme plasma coating greatly reduced total protein adsorption from human blood plasma. As shown in Figure 2, incubation of the tetraglyme-coated sample with 1% plasma resulted in a negligible SPR wavelength shift (0.2 nm, equivalent to 1.3 ng/cm2; see later). After buffer rinsing to remove the loosely bound proteins, the wavelength shift became even smaller and returned to the baseline level. In contrast, incubation of 1% plasma with bare gold resulted in a SPR wavelength shift of 14 nm. After buffer rinsing of this sample, the wavelength shift was still 14 nm (equivalent to 244 ng/cm2), suggesting expð2db =ld Þ Bare gold: R ¼ mðZa Zs Þ½1 expð2da =ld Þ where DR was the SPR wavelength shift; m was the sensitivity factor expressed as the slope of the linear function of wavelength shift versus refractive index change;23 ld was the characteristic decay length calculated from the Maxwell’s equation23 ld ¼ (l/2p)/Re{Z4eff/(Z2eff þ emetal)}1/2 (where l was the light wavelength at the SPR minimum, Zeff was the effective index of refraction, and emetal was the complex dielectric constant of the metal at that wavelength deduced from the paper of Innes et al.29); Za was the average refraction index of plasma protein: 1.6; Zs was the refraction index of the plasma solution; db was the average thickness of deposited tetraglyme films, as measured by AFM. Adsorbed protein thickness da was calculated by substituting values for all other parameters and then da was converted to mass by using the specific volume 0.77cm3/g or density 1.3 g/cm3 of serum albumin.30 As it can be seen from Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a 832 Figure 3. Fg adsorption from 1% human plasma to untreated and tetraglyme plasma treated Biospan, glass, and PE disks. Open bars, untreated samples; solid bars, tetraglyme plasma treated samples. Error bar represents SD. (N ¼ 5, *p < 0.01, Student’s t test). the equation, the wavelength shift will decrease exponentially with the increase of deposited film thickness. If the thickness of deposited polymer film equals the characteristic decay length, the sensitivity would be dropped by a factor of 7. In our case, the thickness of tetraglyme coating (100 nm) was less than 25% of the characteristic decay lengths for all situations; therefore, it did not significantly affect the sensitivity. The detection limit for tetraglyme-coated SPR system was 0.0056 nm in thickness or 0.7 ng/cm2 in mass based on the equations mentioned earlier. Based on the aforementioned equations, the calculated mass of total adsorbed proteins from 1% human plasma on bare gold was 244 ng/cm2, whereas it was 1.3, 4.8, 17.3, and 24.1 ng/cm2 total adsorbed proteins from 1, 10, 50, and 100% plasma, respectively, on tetraglyme-coated gold surfaces. CAO ET AL. Figure 4. vWf adsorption from 1% human plasma to untreated and tetraglyme plasma treated Biospan, glass, and PE disks. Open bars, untreated samples; solid bars, tetraglyme plasma treated samples. Error bar represents SD. (N ¼ 5. *p < 0.01, Student’s t test). Platelet adhesion and procoagulant activity Figure 5 shows that tetraglyme coating of Biospan, glass, or PE greatly decreased platelet adhesion when compared with the uncoated materials (p < 0.01). For tetraglyme-coated samples, the lowest platelet adhesion was on coated PE, while the highest was on coated Biospan. For the uncoated materials, glass had the most adherent platelets, and Biospan had the lowest platelet adhesion. Activation of surface-adherent platelets was assessed by measuring the rate of formation of thrombin in prothrombin solutions incubated with the adherent platelets. Thrombin generation is affected by both the number of surface-adherent platelets as well as the degree of activation of adherent platelets. As shown in Figure 6, all the Protein adsorption from plasma As shown in Figures 3 and 4, Fg and vWf adsorption from 1% plasma were both decreased on tetraglyme plasma coated Biospan, glass, and PE when compared with uncoated controls (p < 0.01). Uncoated PE had the most Fg adsorption (75 ng/cm2), while uncoated glass had the most vWf adsorption (1 ng/cm2). Uncoated Biospan had the least Fg and vWf adsorption, showing that Biospan was more protein-resistant than hydrophobic PE and glass. On the other hand, the chemical nature of the substrates seemed to have little effect on the protein resistance once they were coated with tetraglyme. For example, Fg adsorption was similar on all tetraglyme-coated materials (3 ng/cm2), and vWf adsorption was around 0.2 ng/cm2 for all tetraglyme coatings. These data are consistent with our previous results that showed the high resistance of the plasmadeposited tetraglyme coating to Fg adsorption. Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a Figure 5. Platelet adhesion from washed platelets suspensions to untreated and tetraglyme plasma treated Biospan, glass, and PE disks. Open bars, untreated samples; solid bars, tetraglyme plasma treated samples. Error bar represents SD. (N ¼ 5. *p < 0.01, Student’s t test). IMPROVED HEMOCOMPATIBILITY OF TETRAGLYME PLASMA COATINGS 833 when compared with uncoated glass or FEP (p < 0.01). This showed that tetraglyme coatings activated the intrinsic coagulation cascade less than uncoated controls. Glass had the shortest clotting time, suggesting that it was potent in activating the intrinsic coagulation cascade. Thrombus deposition in an in vitro model of catheterization Figure 6. Thrombin generation by surface adherent platelets to untreated and tetraglyme plasma treated Biospan, glass, and PE disks. Open bars, untreated samples; solid bars, tetraglyme plasma treated samples. Error bar represents SD. (N ¼ 5, *p < 0.01, Student’s t test). tetraglyme-coated samples showed decreased thrombin generation when compared with uncoated samples. PE had the most thrombin generation while Biospan had the least thrombin generation. It is notable that although Biospan had much lower platelet adhesion than glass and PE (Fig. 5), there was not much difference in thrombin generation for Biospan, glass, and PE (Fig. 6). Recalcified plasma clotting times As shown in Figure 7, tetraglyme coating of glass or FEP extended the clotting time of recalcified plasma The importance of fluid dynamics in the thrombotic process, especially disturbed flow that enhances interplatelet collisions and transport of thrombotic components to the biomaterial surface, makes models that incorporate blood flow particularly relevant to the assessment of device thrombosis. Thus, an in vitro model that incorporated disturbed blood flow was utilized to assess the efficacy of tetraglyme-coated catheters in reducing catheter thrombosis.28 As seen in Figure 8, the amount of end-point thrombus (measured as adherent 111In platelets) on the tetraglyme-coated devices was significantly less than the uncoated control (p < 0.05, paired t-test, n ¼ 5), as measured in this bovine in vitro flow model (Fig. 1). SEM examination of the sample corroborated the quantitative data, and showed that the tetraglyme-coated devices had substantially less thrombus on the surface compared with the uncoated control (Fig. 9). DISCUSSION Thrombosis and embolism remain major challenges for medical devices that are used in contact with blood. Therefore, anticoagulant therapy is usually given to patients with cardiovascular implants. However, difficulties occur with the use of anticoagulants (e.g. bleeding episodes). Thus there is still a great need for materials that have reduced reactivity with blood. Since pro- Figure 7. Clotting time of recalcified platelet poor plasma in the presence of different materials. All samples were put in the 24well nontissue culture treated PS plate. ‘‘Polystyrene’’ sample means the well plate only. (N ¼ 6, *p < 0.01, Student’s t test). Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a 834 Figure 8. Thrombus accumulation on surfaces studied in the catheter model. The results are expressed as a percent of the uncoated control. The tetraglyme-coated sample showed a statistically significant reduction in thrombus accumulation (N ¼ 5, *p < 0.05, Student’s t test). tein adsorption onto biomaterials takes place prior to cellular interaction with the surfaces, and is known to promote platelet adhesion and activation, control of the CAO ET AL. amount and composition of the adsorbed protein is a good strategy to improve blood compatibility. Our previous studies have indicated that Fg and vWf are the most important plasma proteins in mediating platelet adhesion.12,18 Thus a surface that could resist the adsorption of Fg and vWf would be expected to have reduced platelet adhesion. On the other hand, reduction of Fg and vWf adsorption should not come at the price of allowing increased adsorption of other platelet adhesive plasma proteins. Adhesive proteins such as Vn and Fn, when present on surfaces in high amounts (e.g., by preadsorption with pure solutions of these proteins), were able to mediate platelet adhesions as well.18,26 Driven by these considerations, we tested whether plasma-deposited tetraglyme coatings previously shown to be highly Fg and vWf resistant would also resist the adsorption of other blood proteins. Results from SPR suggested that plasma-deposited tetraglyme coating greatly reduced total protein adsorption, even when exposed to 100% human blood plasma. Figure 9. SEM images of surfaces in the catheter model. (a,b): Magnification 27; (c,d): Magnification 200. (a,c): Tetraglyme plasma coated PE catheter; (b,d): Uncoated PE catheter. Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a IMPROVED HEMOCOMPATIBILITY OF TETRAGLYME PLASMA COATINGS The reason why plasma-deposited tetraglyme coatings are highly protein-resistant is not fully understood. We believe the ability of plasma-deposited tetraglyme coatings to tightly structure water is the reason that they exhibit strong inhibition of protein adsorption. This ability is thought to be associated with two factors. First, the ether linkages in the tetraglyme coatings bind and tightly structure the water into ‘‘ice-like’’ forms. Second, the plasma deposition process creates relatively short ethylene oxide (EO) chains that are able to fill in and cover the surface and achieve a high surface density of EO units. In addition, the conformation of the EO chains in the tetraglyme coating may also play a role in inhibiting protein adsorption. Li et al.31 found that there was an optimal packing density of oligo ethylene glycol self-assembled monolayers (OEG-SAMs) to resist protein adsorption. For the most tightly packed OEG-SAMs, protein adsorption was higher than on less densely packed OEG surfaces, apparently because the high-packing made it more difficult for the access of water molecules to the OEG. Since the tetraglyme chains produced from the deposition process do not self assemble, but are more likely arranged in a random orientation to the surface with many chains parallel to the surface plane, the high surface density of EO chains is unlikely to limit the accessibility to water molecules. AFM force measurements by Bremmell et al.19 showed a steric repulsive nature of similar PEO-like coatings. Thus it is possible that tetraglyme plasma coatings exhibited similar steric–entropic barrier nature. Antonsen and Hoffman32 and Tanaka et al.33–35 showed that ‘‘freezing bound water’’ appeared to be critical in making polyethylene glycol and poly(2methoxyethyl acrylate) (PMEA) repel protein. The EO functional group in the tetraglyme glow discharge-deposited coating may cause a similar freezing bound water effect and make it highly protein resistant. If protein adsorption is driven by the entropic increase associated with ‘‘melting’’ by displacement of surface-bound water, and if EO units create tightly bound ‘‘frozen’’ water, then the EG units may prevent melting, thus eliminating the driving force for protein adsorption. Tanaka et al.33–35 also suggested that adsorbed proteins on PMEA retained a more native conformation than when they were adsorbed on poly 2-hydroxyethylmethacrylate (PHEMA). Even though the hydrophobicity of PMEA and PHEMA are similar, the Fg bound to PMEA was not only lower in amount, but was less active in binding platelets, and retained more of its native conformation. On the surface of tetraglyme coatings exhibiting very low Fg adsorption, the tight binding of water may allow adsorbed Fg to be in a more native conformation with consequent lower platelet reactivity. However, investigation of the conformation and orientation of adsorbed proteins on tetraglyme coatings, for example, by probing the availability of platelet binding epitopes36 is difficult, because the total amount of adsorbed protein was very low. 835 Although the tetraglyme plasma coating greatly inhibited total protein adsorption, there were still some platelets adherent to the surface. Whether these sparsely adherent platelets are bound to the residual adsorbed Fg or vWf, or possibly by other platelet adhesive proteins such as Vn and Fn, is still not clear. However, it seems mostly likely that the residual platelet adhesion is due to the small residual Fg adsorption, given the extreme sensitivity of platelets to adsorbed Fg. Further studies to identify all the proteins adsorbed from plasma onto tetraglyme glow discharge deposited surfaces may provide more information and insights about the relationship between protein adsorption and blood cells attachment and activation.37 For example, in a recent report by Elbert and coworkers,38 it has been shown that a plasma protein named ‘‘serum amyloid P’’ was found to be critical in mediating monocyte adhesion, which was not detected before by other techniques. On the other hand, assuming that all the platelet adhesion related proteins have been identified, it is still a challenge to make a material that resists the adsorption of all the adhesive proteins. This requires a further understanding of the mechanism of protein–material interactions. We believe that the residual adsorption to our ultralow protein adsorption tetraglyme coatings is due to two factors. First, there are probably small areas of the substrate that are not well covered by the plasma coating. Second, the small content of hydrophobic fragments on the tetraglyme coating that are inevitably generated during the plasma deposition process may also adsorb proteins. Although it is known that platelet adhesion is mainly mediated by adsorbed Fg under static situations, a comparison of Figures 3 and 5 showed that the highest platelet adhesion did not occur on the surface that had the highest Fg adsorption (uncoated PE), but on uncoated glass. This was probably due to the fact that the potency of adsorbed Fg in supporting platelet adhesion depends on both the surface it is adsorbed to and the adsorbed amount, which is associated with variable exposure of the platelet binding epitope that can be measured with monoclonal antibodies.36 Platelets adherent to Biospan were in a more procoagulant state than platelets on PE and glass. The rates of thrombin formation on tetraglyme-coated Biospan, glass, or PE in the presence of adhered platelets were much lower than on the uncoated surfaces (Fig. 6), but the reduction in thrombin formation by the tetraglyme coating was not as large as the reduction in platelet adhesion. These results differed from that from previous thrombin formation studies we did with tetraglymecoated tubes,16 in which the thrombin formation was much lower than reported here. We believe that the difference may be due to the use of a PS well to contain the flat surfaces in the current studies, whereas in the tube studies the only surface present was the inside wall of the tetraglyme-coated tube. Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a 836 In the clotting time study, glass had the shortest clotting time when compared with other surfaces. This was expected, as it is well known that surfaces such as glass initiate contact activation by adsorbing prekallikrein (PK), high-molecular-weight kininogen (HMWK), factor XI, and factor XII. In contrast, the tetraglyme coatings, comprised mostly of short ethylene glycol chains, are highly protein-resistant and they probably inhibit the adsorption of PK, HMWK, factor XI, and factor XII, thus decreasing the activation of the intrinsic coagulation cascade. PS had a longer clotting time than FEP. This is probably due to the differences in their adsorption of clotting factors. In vitro models that incorporate disturbed blood flow, such as the one used in the current study, provide a more realistic assessment of device-induced thrombosis. The conditions (e.g., blood flow, hemostasis) in in vitro models are relatively more controlled than in vivo models. This allows the experiment to be focused on a specific variable (e.g. surface chemistry), with other relevant parameters (e.g. disturbed flow) being present but remaining relatively constant. However, it is important to note that there are differences between in vitro and in vivo models that make it difficult to directly extrapolate in vitro results to in vivo situations. Items that are present in vivo but absent in vitro include (i) a more dynamic and comprehensive hemostatic system (e.g. thrombolytic pathway) and (ii) less activated blood components (e.g. platelets and clotting factors). Further, in vitro models require at least some anticoagulation of blood to prevent clotting of the entire pool of blood during the experiment, while this is not necessarily in the case of in vivo situations. Also, the in vitro system allows for the assessment of acute-phase thrombosis (*hours), but does not provide any information on chronic thrombogenicity (*days or *months). Thus, while the thromboresistance exhibited by the tetraglyme-coated catheters under disturbed flow conditions in vitro is encouraging, evaluation in long-term implants in vivo is required to reveal the ability of tetraglyme coatings to resist chronic thrombus formation. In studies published using a long-term ex vivo shunt model of platelet interaction, platelet consumption was noted to increase with increasing equilibrium water content of radiation-grafted hydrogels.39,40 PEO-like polymers are associated with high water contents. However, we believe our present results are not contradictory to this previously published water content trend. First, our tetraglyme coating is so tightly crosslinked that it exhibits little swelling and, in fact, appears neither highly hydrophobic nor hydrophilic by contact angle measurement.41,42 A study by Bremmell et al.19 by AFM force measurements suggested that their similar PEO-like plasma coatings possessed a steric repulsion force indicative of hydrogel compression. Thus tetraglyme coatings may exhibit the steric repulsive nature to some extent. Second, PEO has no hydrogen bond doJournal of Biomedical Materials Research Part A DOI 10.1002/jbm.a CAO ET AL. nor groups, whereas all the polymers assessed in the report by Hanson et al.39,40 did contain hydrogen bond donor groups. Hydrogen bond donor groups were postulated to be responsible for blood interactions.43,44 We thank Winston Ciridon for assistance with plasma deposition, Jim Hull for the AFM study, and Dr. Steve Golledge, Dr. Lara Gamble, and Dr. Dan Graham for ESCA operation and analysis. The SPR, ESCA, and AFM experiments were performed at the National ESCA and Surface Analysis Center for Biomedical Problems. We thank the many individuals who donated blood for our studies. References 1. Ratner BD. Blood compatibility—A perspective. J Biomater Sci Polym Ed 2000;11:1107–1119. 2. Hoffman AS. Nonfouling surface technologies. J Biomater Sci Polym Ed 1999;10:1011–1014. 3. Lee JH, Lee HB, Andrade JD. Blood compatibility of polyethylene oxide surfaces. Prog Polym Sci 1999;20:1043–1079. 4. Sefton MV. Nonthrombogenic treatments and strategies. In: Ratner BD, Schoen FJ, Hoffman AS, Lemons JE, editors. Biomaterials Science: An introduction to Materials in Medicine. New York: Elsevier; 2004. p 456–470. 5. Lewis AL. Phosphorylcholine-based polymers and their use in the prevention of biofouling. Colloids Surf B 2000;18:261–275. 6. Park K, Shim HS, Dewanjee MK, Eigler NL. In vitro and in vivo studies of PEO-grafted blood-contacting cardiovascular prostheses. J Biomater Sci Polymer Ed 2000;11:1121–1134. 7. Kim JH, Song M, Roh HW, Shin YC, Kim SC. The in vitro blood compatibility of poly(ethylene oxide)-grafted polyurethane/polystyrene interpenetrating polymer networks. J Biomater Sci Polymer Ed 2000;11:197–216. 8. Ruiz L, Fine E, Voros J, Makohliso SA, Leonard D, Johnston DS, Textor M, Mathieu HJ. Phosphorylcholine-containing polyurethanes for the control of protein adsorption. J Biomater Sci Polym Ed 1999;10:931–955. 9. Horbett TA. Principles underlying the role of adsorbed plasma proteins in blood interactions with foreign materials. Cardiovasc Pathol 1993;2:137s–148s. 10. Hanson SR. Blood coagulation and blood-materials interactions. In: Ratner BD, Schoen FJ, Hoffman AS, Lemons JE, editors. Biomaterials Science: An introduction to Materials in Medicine. New York: Elsevier; 2004. p 332–338. 11. Tsai W-B, Grunkemeier JM, Horbett TA. Human plasma fibrinogen adsorption and platelet adhesion to polystyrene. J Biomed Mater Res 1999;44:130–139. 12. Kwak D, Wu Y, Horbett TA. Fibrinogen and von Willerbrand’s factor adsorption are both required for platelet adhesion from sheared suspensions to polyethylene preadsorbed with blood plasma. J Biomed Mater Res A 2005;74:69–83. 13. López GP, Ratner BD, Tidwell CD, Haycox CL, Rapoza RJ, Horbett TA. Glow discharge plasma deposition of tetraethylene glycol dimethyl ether for fouling-resistant biomaterial surfaces. J Biomed Mater Res 1992;26:415–439. 14. López GP, Ratner BD. Molecular adsorption and the chemistry of plasma-deposited thin organic films: Deposition of oligomers of ethylene glycol. Plasmas Polym 1996;1:127–151. 15. Shen MC, Pan YV, Wagner MS, Hauch KD, Castner DG, Ratner BD, Horbett TA. Inhibition of monocyte adhesion and fibrinogen adsorption on glow discharge plasma deposited tetraethylene glycol dimethyl ether. J Biomater Sci Polym Ed 2001;12:961–978. 16. Cao L, Sukavaneshvar S, Ratner BD, Horbett TA. Glow discharge plasma treatment of polyethylene tubing with tetraglyme results IMPROVED HEMOCOMPATIBILITY OF TETRAGLYME PLASMA COATINGS 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. in ultralow fibrinogen adsorption and greatly reduced platelet adhesion. J Biomed Mater Res. Forthcoming. Horbett TA. Mass-action effects on competitive adsorption of fibrinogen from hemoglobin-solutions and from plasma. Thromb Haemost 1984;51:174–181. Tsai W-B, Grunkemeier JM, McFarland CD, Horbett TA. Platelet adhesion to polystyrene-based surfaces preadsorbed with plasma selectively depleted in fibrinogen, fibronectin, vitronectin, or von Willebrand’s factor. J Biomed Mater Res 2001;60:348–359. Bremmell KE, Kingshott P, Ademovic Z, Winther-Jensen B, Griesser HJ. Colloid probe AFM investigation of interactions between fibrinogen and PEG-like plasma polymer surfaces. Langmuir 2006;22:313–318. Mrksich M, Sigal GB, Whitesides GM. Surface-plasmon resonance permits in-situ measurement of protein adsorption on selfassembled monolayers of alkanethiolates on gold. Langmuir 1995; 11:4383–4385. Sigal GB, Bamdad C, Barberis A, Strominger J, Whitesides GM. A self-assembled monolayer for the binding and study of histidine tagged proteins by surface plasmon resonance. Anal Chem 1996;68:490–497. Shen MC, Martinson L, Wagner MS, Castner DG, Ratner BD, Horbett TA. PEO-like plasma polymerized tetraglyme surface interactions with leukocytes and proteins: In vitro and in vivo studies. J Biomater Sci Polym Ed 2002;13(4):367–390. Jung LS, Campbell CT, Chinowsky TM, Mar MN, Yee SS. Quantitative interpretation of the response of surface plasmon resonance sensors to adsorbed films. Langmuir 1998;14:5636– 5648. Wu Y, Simonovsky FI, Ratner BD, Horbett TA. The role of adsorbed fibrinogen in platelet adhesion to polyurethane surfaces: A comparison of surface hydrophobicity, protein adsorption, monoclonal antibody binding and platelet adhesion. J Biomed Mater Res A 2005;74:722–738. Grunkemeier JM, Tsai W-B, Alexander MR, Castner DG, Horbett TA. Platelet adhesion and procoagulant activity induced by contact with radio frequency glow discharge polymers: Roles of adsorbed fibrinogen and vWf. J Biomed Mater Res 2000;51:669– 679. Grunkemeier JM, Tsai W-B, McFartland CD, Horbett TA. The effect of adsorbed fibrinogen, fibronectin, von Willebrand factor and vitronectin on the procoagulant state of adherent platelets. Biomaterials 2000;21:2243–2252. Grunkemeier JM, Tsai W-B, Horbett TA. Hemocompatibility of treated polystyrene substrates: Effect of contact activation, platelet adhesion and procoagulant state of adherent platelets. J Biomed Mater Res 1998;41:657–670. Sukavaneshvar S, Rosa GM, Solen KA. Enhancement of stentinduced thromboembolism by residual stenoses: Contribution of hemodynamics. Ann Biomed Eng 2000;28:182–193. Innes RA, Sambles JR. Optical characterization of gold using surface plasmon-polaritons. J Phys F: Met Phys 1987;17:277– 287. 837 30. Sjolander S, Urbaniczky C. Integrated fluid handling-system for biomolecular interaction analysis. Anal Chem 1991;63:2338–2345. 31. Li L, Chen S, Zheng J, Ratner BD, Jiang SY. Protein adsorption on oligo(ethylene glycol)-terminated alkanethiolate self-assembled monolayers: The molecular basis for nonfouling behavior. J Phys Chem B 2005;109:2934–2941. 32. Antonsen KP, Hoffman AS. Water structure of PEG solutions by differential scanning calorimetry measurements. In: Harris JM, editor. Poly(ethylene glycol) Chemistry: Biotechnical and Biomedical Applications (Topics in Applied Chemistry). New York: Plenum; 1992. 33. Tanaka M, Motomura T, Kawada M, Anzai T, Kasori Y, Shiroya T, Shimura K, Onishi M, Mochizuki A. Blood compatible aspects of poly(2-methoxyethylacrylate) (PMEA) relationship between protein adsorption and platelet adhesion on PMEA surface. Biomaterials 2000;21:1471–1481. 34. Tanaka M, Mochizuki A. Effect of water structure on blood compatibility-thermal analysis of water in poly(meth)acrylate. J Biomed Mater Res A 2004;68:684–695. 35. Tanaka M, Mochizuki A, Ishii N, Motomura T, Hatakeyama T. Study of blood compatibility with poly(2-methoxyethyl acrylate). Relationship between water structure and platelet compatibility in poly(2-methoxyethylacrylate-co-2-hydroxyethylmethacrylate). Biomacromolecules 2002;3:36–41. 36. Tsai W-B, Grunkemeier JM, Horbett TA. Variations in the ability of adsorbed fibrinogen to mediate platelet adhesion to polystyrene-based materials: A multivariate statistical analysis of antibody binding to the platelet binding sites of fibrinogen. J Biomed Mater Res A 2003;67:1255–1268. 37. Horbett TA. Adsorption of proteins from plasma to a series of hydrophilic–hydrophobic co-polymers. II. Compositional analysis with the prelabeled protein technique. J Biomed Mater Res 1981;15:673–695. 38. Kim J-K, Scott EA, Elbert DL. Proteomic analysis of protein adsorption: Serum amyloid P adsorbs to materials and promotes leukocyte adhesion. J Biomed Mater Res A 2005;75:199–209. 39. Hanson SR, Harker LA, Ratner BD, Hoffman AS. In vivo evaluation of artificial surfaces with a nonhuman primate model of arterial thrombosis. J Lab Clin Med 1980;95:289–304. 40. Hanson SR, Harker LA, Ratner BD, Hoffman AS. Factors influencing platelet consumption by polyacrylamide hydrogels. Ann Biomed Eng 1979;7:357–367. 41. Johnston EE, Bryers JD, Ratner BD. Plasma deposition and surface characterization of oligoglyme, dioxane, and crown ether nonfouling films. Langmuir 2005;21:870–881. 42. Johnston EE. Surface and biological properties of biofoulingresistant, poly(ethylene oxide)-like plasma deposited films. Ph.D. Dissertation, University of Washington, 1997. 43. Ostuni E, Chapman RG, Holmlin RE, Takayama S, Whitesides GM. A survey of structure–property relationships of surfaces that resist the adsorption of protein. Langmuir 2001;17:5605–5620. 44. Merrill EW. Distinctions and correspondences among surfaces contacting blood. Ann NY Acad Sci 1987;516:196–203. Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a