Extracorporeal membrane oxygenation (ECMO) has been available for clinical cardiopulmonary support since the late 1970s. Over the last 35 years, technology has improved, critical care has evolved, and patient disease cohorts have been identified where ECMO is routinely applied in prenatal, pediatric, and adult populations. Since the H1N1 epidemic of 2007, application of ECMO in adults with severe respiratory failure has greatly increased.1 Application of ECMO in critical care cardiopulmonary support is trending to longer utilization (weeks to months) and ambulation, especially regarding respiratory failure as a bridge to recovery or transplant.2,3 Most recently, at the 2015 Annual Meeting of the American Association of Thoracic Surgeons, guidelines were discussed recommending the use of ambulatory ECMO as a bridge to lung transplant and for extracorporeal lung support.
Current oxygenators use polymeric hollow fiber membranes (HFMs), many of which are based on either polypropylene (PP) or polymethylpentene (PMP).4–7 Surface modification of HFMs to form coated (or composite) membranes is an intense area of study and has been the subject of several reviews. Most, if not all, of these coating strategies would perform best when the exterior membrane surface has no contamination or extraneous materials. The presence of a surface contaminant could potentially interfere with covalent bonds, surface associations, or wetting between the HFM exterior surface and the applied material.
HFMs are usually manufactured using thermally induced phase separation processes. The phase-separating agent typically dissolves in the polymer at extrusion temperatures but separates from the polymer at lower temperatures, creating porosity within the hollow fiber walls.8 In the cases of commercial PP and PMP hollow fibers, the phase-separating agent can be removed from the membrane by extraction with solvents such as 1,1,1-trichloroethane or isopropyl alcohol.9,10
Residual oil on the surface of fibers may cause problems with postsynthesis coatings, including poor interfacial adhesion or incomplete coverage of all fiber surfaces. The delamination of such coatings within blood oxygenators has the potential to severely injure or kill a patient.
Leaching of the phase separation agent into blood during oxygenator use may cause some health problems. Soybean oil, a commonly used phase-separating agent, is known to be toxic to leukocytes. Linoleic acid, a major component of soybean oil, has been shown to cause apoptosis of human T-cells.11 Linoleic acid causes mitochondrial depolarization and the generation of reactive ion species within lymphocytes.12 Oleic acid, a component of soybean oil, induced apoptosis by activation of capsase 3.12 If leached from the large blood contacting surface area of modern blood oxygenators, soybean oil may cause suppression of the immune response, promoting infection. Consequently, the presence and amount of any substance leaching from these devices should be determined. Food-grade mineral oil has been approved by the Food and Drug Administration for direct or indirect contact with food and is generally recognized as safe. Although it might be tolerated by the body, it could still impact postsynthesis coating processes. The objectives of this work were to 1) leach any residual agents from commercial HFMs, 2) identify the general category of the leached material, and 3) evaluate the quality of a surface-initiated atom transfer radical polymerization (s-ATRP) coating on unleached and leached hollow fibers with respect to complete surface coverage and gas leakage rate.
Materials and Methods
X30-240 PP HFM and Oxyplus PMP HFM were acquired from Celgard. Chromosolv high-pressure liquid chromatography (HPLC) grade acetone with purity of greater than 99.9% was purchased from Sigma-Aldrich (Billerica, MA). Poly(ethylene glycol) methacrylate (PEGMA) with molecular weight of 360 was purchased from Sigma-Aldrich (Billerica, MA), passed through inhibitor removal columns (Sigma), and stored in a refrigerator before use. NOCHROMIX was used to clean all glassware used in this study and was purchased from Sigma-Aldrich (Billerica, MA). Bromoisobutyryl bromide, 99.9% anhydrous tetrahydrofuran, triethylamine, copper (I) chloride, copper (II) chloride, 2-2′ bipyridine, and technical grade sulfuric acid were purchased from Sigma-Aldrich (Billerica, MA). The castable resin used for embedding the fibers in the cylinder was WC-781 from BJB Industries.
A Hitachi Primaide system with a Kinetex C-18 column (5 µm particles, 100 Å pores, 150 × 4.60 mm; Phenomenex, Torrance, CA) was used to perform all HPLC measurements. The column size was 150 × 4.60 mm. A ChemGlass Soxhlet apparatus and condenser was for all extractions. Figure 1 is a sketch of the pressure cell used for determining the pore size of the HFMs before and after s-ATRP coating. The pressure cell consisted of a steel cylinder with diameter of 60 mm and length of 215 mm. A 12.7 mm section of steel was welded onto one end of the cylinder, and a ½ inch national pipe thread taper threaded hole was created. On the opposing end of the cylinder, a 127 mm diameter ring of thickness 12.7 mm was welded onto the 60 mm cylinder. A mating surface was created to accommodate the fiber samples embedded in a castable resin. The upper portion of the pressure cell was created in the same manner, however, with a 37 mm long cylinder.
Extraction of residual oils. Approximately 2 gm of hollow fibers, either PP or PMP, were wound around two wooden dowels of 0.5 inch diameter spaced at a distance of 30 inches. This amounts to 0.076 km of fiber or about 1% of the amount found in an adult-sized oxygenator. Sections of clean tantalum wire with length of two inches were weighed and then wrapped around each end of the fiber to create a bundle. These bundles were then weighed (compensating for the tantalum wire weight). Three bundles of each hollow fiber type were extracted to determine the weight fraction of residual oils.
The Soxhlet extraction apparatus and all glassware were cleaned by the following procedure. The glassware was immersed in a potassium hydroxide/isopropyl alcohol bath for a period of 24 hours. The glassware was then rinsed with deionized (DI) water and cleaned with a NOCHROMIX/sulfuric acid solution for 2 hours followed by rinsing with DI water and storage in an oven at 90°C until use.
Each HFM bundle was individually placed into a Soxhlet extractor apparatus with a 500 ml round bottom flask as the solvent reservoir (500 ml solvent was used). CHROMOSOLV HPLC grade acetone (Sigma-Aldrich, Billerica, MA) with purity greater than 99.9% was used for all extractions. The extractor was run for 36 continuous hours. The key objective of this method was to remove all residual oils, so the leaching rate was not assessed. The acetone needed to be removed from the leachate to detect residual oils using HPLC.
HPLC analysis of extracted residue. 20 ml of acetone solvent used to extract the fibers was placed in a flask at 80°C to evaporate the solvent. Afterward, 2 ml of pure acetone was added to the flask to redissolve the residue. Afterward, this 2 ml of the remaining solvent was analyzed by HPLC. Samples (50 µl) were injected into the column with a gradient mobile phase and flow rate of 0.5 ml/min, with absorbance measurements of column effluent made at 254 nm.
Although previous literature reports used a water-to-acetonitrile solvent gradient to elute fatty acid peaks,13 we found that this gradient did not completely elute all of the unknown residue in the C18 column. For this reason, a second solvent gradient was added, acetonitrile-to-ternary solution, which contained acetonitrile, ethanol, and hexane.14Table 1 shows the mobile phase composition versus time for the HPLC runs. The initial elution solvent is very hydrophilic, so hydrophobic solutes will adsorb onto the packing. The adsorption and release of solutes from the C-18 column can be interpreted using Hansen solubility parameters.
S-ATRP of hollow fiber membranes. Using the protocol developed by Yao et al.,15 a PEGMA coating was polymerized on the surface of isolated hollow fibers to a thickness of approximately 100 nm.
Gas leakage rate and average pore size determination. Each fiber sample was embedded into the pressure cell, and the porosity of the PP hollow fibers was then tested using ASTM Standard E128,16 modified for the measurement of pore size and distribution on the surface of HFMs. Deionized water was used as the fluid for detecting bubble formation due to gas leaking through membrane pores. Nitrogen was then applied to the bore side of the fiber at increasing pressures until the first bubble broke from the surface of the fiber. This pressure was noted as the bubble point.
High-Pressure Liquid Chromatography Chromatograms for Soybean and Mineral Oils
Table 2 lists Hansen solubility parameter values for the individual solvents, solvent mixtures, and solutes in the HPLC separations. The starting solvent in the HPLC separations is water mixed with 0.1% trifluoroacetic acid, a strongly hydrogen-bonding solution. Since the stationary phase of the column packing is functionalized with ligands containing 18 carbons, either mineral oil components or soybean oil components should partition to the stationary phase. As the solvent phase composition changes during the HPLC separations, the continuous phase, Hansen solubility parameters will change. When there is an adequate match between the solute and the solvent properties, the solute will desorb from the packing into the solvent, flow through the column, and be detected at the outlet.
For purposes of estimating differences between the Hansen solubility parameters of the target solute and the solvent mixture, oleic acid was used to represent soybean oil and decane was used to represent mineral oil. These Hansen solubility parameter differences are plotted for both the oleic acid–solvent and decane–solvent pairs versus elution time (Figure 2). All of the solvent–solute pairs have small differences between the disperse solubility parameters; the solvent changes do not affect the differences in Hansen dispersive components. During the change from solvent 1 to solvent 2, the differences between hydrogen-bonding values for the solute–solvent pair decrease significantly although there is a 10% increase in the differences between the polar components of the solute–solvent pair. During the change from solvent 2 to solvent 3, the differences between the polar component of both pairs decrease, and the differences between the hydrogen-bonding component of both pairs increase.
Figure 3 shows the HPLC chromatogram for soybean oil in an acetone carrier. There is an initial solvent (acetone) peak at 5 minutes retention time. A few components are eluted near 50 minutes, but most of the sample is eluted between 140 and 190 minutes. This elution range corresponds to the region over which the differences between the polar components decrease significantly. Soybean oil contains seven or more long chain fatty acids.13Figure 3 indicates the presence of multiple peaks of hydrophobic components.
Figure 4 shows the elution curve for mineral oil, which appears between 60 and 80 minutes, followed by a slow release as the third solvent is introduced (>120 minutes). The first release coincides with the difference in the hydrogen-bonding components being reduced below 20 MPa1/2. Material released beyond 120 minutes coincides with a slow reduction in the difference between the polar components. Figures 3 and 4 show that soybean oil and mineral oil can be distinguished by this particular HPLC protocol.
Residue Extracted from Hollow Fiber Membranes
The leachate from Soxhlet extraction was concentrated to enhance the signal from HPLC analysis. Concentration was accomplished by boiling the acetone (~56°C is the boiling point). Both soybean oil and decane have much higher boiling points (177 and 174°C, respectively), so little of the solutes should have evaporated during the process. After evaporation of the acetone solvent, brown liquids remained. The HPLC chromatograms of the concentrated leachate residues are shown in Figures 5 and 6. Multiple peaks that elute as the mobile phase becomes more hydrophobic were observed. These multiple peaks, also observed in mineral oil and soybean oil (Figures 3 and 4), are determined to be hydrophobic oils. Both materials used for HFM blood oxygenators leached significant amounts of oily residues.
The weight percent (wt %) weight loss was 2.50 ± 0.36% for PP and 7.13 ± 5.47% for PMP after 36 hours of extraction with acetone. Oils retained from HFM manufacturing processes may interfere with a number of postproduction treatments, such as surface functionalizations and coatings.
Poly(Ethylene Glycol) Methacrylate Coatings on Hollow Fiber Membranes
Scanning electron microscopy. Both received and leached HFMs were modified with s-ATRP coatings. The procedure was designed to make uniform coatings of ~100 nm thickness on the HFMs. Coated fibers were examined using scanning electron microscopy to evaluate the uniformity of the surface coatings. Fibers that were not leached before coating showed large areas, on the order of 100 µm2, over which no coating was detected or the coating was incomplete. Polypropylene and PMP hollow fibers that leached with acetone for 36 hours followed by functionalization by the s-ATRP process had continuous, intact coating coverages.
Figure 7A–C shows exterior surface images of unleached PP hollow fibers before and after s-ATRP coatings. Figure 7A displays the virgin fiber as received. Fibers that were not cleaned displayed an incomplete coating (Figure 7B). In these areas, the surface appears similar to that of the as-received fibers, with a series of pore openings present throughout the surface. Figure 7C shows an area on which PEGMA coating adhered; there are still some apparently open, but smaller, surface pores. Figure 8 shows the PEGMA coating on cleaned PP hollow fibers. Scanning electron microscopy evaluation of this sample showed no areas similar to those in Figure 7B or C; the PEGMA-coated surface appeared to be continuous and complete.
Bubble point testing. The presence of a continuous coating across the entire surface area of the PP HFM was further indicated by the pore size measurement by bubble point testing as shown in Table 3. The maximum pressure achieved in the pressure cell was 160 psi; at this point, no bubbles could be observed for fibers cleaned before coating deposition. The estimated maximum pore size for this pressure was 0.26 µm, although the actual maximum pore size must be smaller than this value. Therefore, the maximum pore size has decreased by at least 85% when residual oils are removed by leaching.
The surface tension of the PP HFM may be altered by the presence of soybean oil or mineral oil, but it will not be altered much. The surface tension of PP is 30.1 dynes/cm (mN/m), whereas the values for soybean oil and mineral oil are 27.6 and 29.8 dynes/cm, respectively. Assuming that the surface tension of the unleached HFM is modeled by the rule of mixtures, the presence of the oils would decrease the HFM surface tension value by no more than 0.6% on the basis of a maximum loading of 7.5% residual oil. The effective pore size measurement using the bubble point pressure is not expected to be affected significantly by the presence of the oils.
Commercial PMP and PP HFMs, both commonly used in blood oxygenation, may contain significant amounts of an oily contaminant. Extensive Soxhlet extraction of these fibers with acetone caused them to lose between 2.5% and 7.5% of their initial weight. The HPLC chromatograms of the oily contaminants were compared with those of neat soybean oil and neat mineral oil.
Soybean oil showed two sets of elution peaks. The first set appears at elution times of ~40 and ~60 minutes. The two elution peaks are very sharp, suggesting that these might represent very specific compounds. During this time period, the hydrogen bonding difference between the solutes (decane or oleic acid) and the elution mixture is decreasing significantly. The second set of elution peaks appears over the time range of 140–190 minutes. During this time period, the polar component values of the solutes and elution mixtures solubility parameters are becoming more closely matched. Furthermore, there are a number of peaks, suggesting that multiple compounds are being released from the column, such as the many fatty acids that comprise soybean oil.
High-pressure liquid chromatography chromatograms of mineral oil showed two sets of elution peaks. The lower set, released over the time range of 40–90 minutes, showed multiple component peaks, consistent with the many alkane chain components of mineral oil. At higher elution times (t > 140 minutes), there is a slow change in the absorbance amplitude, but no definitive peaks.
The chromatograms of the residue leached from both PMP and PP HMFs showed two sets of components. The PMP elution curve appears to be more similar to that of mineral oil than soybean oil; it shows multiple peaks in the range of 40–90 minutes, with a slow, broad increase after 120 minutes. The chromatogram for the PP leachate shows lower levels of contaminants. However, it also shows a few sharp peaks at 60 and 70 minutes, followed by a gradual increase for times longer than 120 minutes. It appears that the residue leached from both of these hollow fibers is more similar to mineral oil than soybean oil. Because the fabrication protocols for both of these commercial hollow fibers are not generally reported, we were unable to ascertain the phase-separating agent used for the production of either fiber type. However, both types of fiber release low-molecular-weight, oily residue on long-term leaching. This could be caused by insufficient cleaning or the use of a polar solvent that unintentionally leaves the nonpolar components of the cosolvent.
There are likely to be a number of organic solvents that would be effective for leaching residual oils from HFMs made by phase-inversion technologies. Because the most likely candidates for the leachable oil were soybean oil and mineral oil, the extraction solvent was chosen so that it could be used for either nonpolar (mineral oil) or polar (soybean oil) agents. Solvents with boiling points much lower than those of the leachable oils would be preferred as they could be recovered and recycled using simple evaporation or distillation technologies with modest energy costs. It is convenient to perform the leaching process at the normal boiling point of the solvent at atmospheric pressure for similar reasons.
Removal of this residue was found to be critical for the fabrication of a defect-free continuous coating on the entirety of the fiber’s surface. Without extensive cleaning, s-ATRP derived polymer coatings left exposed surfaces on the order of 100 µm2. Surface-initiated atom transfer radical polymerization coating fabrication performed on leached hollow fibers produced a continuous coating. The PEG coating also reduced the average pore size of the hollow fibers. Coated fibers had an estimated maximum pore size of less than 0.26 µm, whereas uncoated fibers had a maximum pore size of 1.67 µm.
The objective of this leaching process was to prepare the HFM membrane for coating processes that could completely cover the open pores present on the exterior surfaces. A submicron coating on the exterior surface would be sufficient to prevent gas and liquid leakage into and out of the HFM, while adding only modest diffusion resistance to gas exchange. Depending on the membrane, solvent, and leaching conditions, there may be modest changes in membrane morphology.
Other HFM membranes should be leachable by similar methods. In general, these are semicrystalline polymers for which residual oils would be present primarily in the amorphous regions of the membranes. Solvents would be selected on the basis of their miscibility with the residual oil and their ability to swell the amorphous regions of the membrane. For example, poor solvents can swell but not dissolve the amorphous polymer. The detailed HFM morphology will affect the rate at which leaching can be done, e.g., the characteristic diameter of polymer structures in contact with liquid solvent might be a rate-controlling factor.
It is not clear what the effects of a leachable agent like mineral oil might be on the performance of HFMs in contact with blood. This would be a topic for further evaluation.
These data show that leachable residue is present in two commercial HFM. A covalently bound coating applied to unleached fibers left a number of open pores and did not form uniform coatings across the entire exterior fiber surface. By contrast, the same coating applied to leached fibers did form uniform coatings and decreased the estimated maximum pore size by over a factor of five. The leachable residue appears to be more similar to mineral oil than soybean oil; both oils have been reported as phase separation agents in the manufacture of hollow fibers.
1. Noah MA, Peek GJ, Finney SJ, et al.Referral to an extracorporeal membrane oxygenation
center and mortality among patients with severe 2009 influenza A (H1N1).JAMA201130616591668
2. Hoopes CW, Kukreja J, Golden J, Davenport DL, Diaz-Guzman E, Zwischenberger JBExtracorporeal membrane oxygenation as a bridge to pulmonary transplantation.J Thorac Cardiovasc Surg20131458627; discussion 867
3. Diaz-Guzman E, Hoopes CW, Zwischenberger JBThe evolution of extracorporeal life support as a bridge to lung transplantation.ASAIO J201359310
4. Kang G-D, Cao Y-MApplication and modification of poly(vinylidene fluoride) (PVDF) membranes—a review.J Membr Sci2014463145165
5. Lau WJ, Ismail AF, Misdan N, Kassim MAA recent progress in thin film composite membrane: A review.Desalination2012287190199
6. Rana D, Matsuura TSurface modifications for antifouling membranes.Chem Rev201011024482471
7. Khulbe KC, Feng C, Matsuura TThe art of surface modification of synthetic polymeric membranes.J Appl Polym Sci2010115855895
8. Lloyd DR, Kinzer KE, Tseng HSMicroporous membrane formation via thermally induced phase separation. I. Solid-liquid phase separation.J Membr Sci199052239261
9. Matsuyama H, Kim M-M, Lloyd DREffect of extraction and drying on the structure of microporous polyethylene membranes prepared via thermally induced phase separation.J Membr Sci2002204413419
10. Kim J-J, Hwang JR, Kim UY, Kim SSOperation parameters of melt spinning of polypropylene hollow fiber membranes.J Membr Sci19951082536
11. Cury-Boaventura MF, Pompéia C, Curi RComparative toxicity of oleic acid and linoleic acid on Jurkat cells.Clin Nutr200423721732
12. Cury-Boaventura MF, Gorjão R, de Lima TM, Newsholme P, Curi RComparative toxicity of oleic and linoleic acid on human lymphocytes.Life Sci20067814481456
13. Jaselskis B, Stemm NL, Johnston WDDetermination of the fatty-acid composition of soybean oil by high-pressure liquid chromatography.Talanta1982295456
14. Herslöf B, Kindmark GHPLC of triglycerides with gradient elution and mass detection.Lipids198520783790
15. Yao F, Fu G-D, Zhao J, Kang E-T, Neoh KGAntibacterial effect of surface-functionalized polypropylene hollow fiber membrane from surface-initiated atom transfer radical polymerization.J Membr Sci2008319149157
16. ASTM International Subcommittee E41.01 on Laboratory Ware and Supplies: ASTM E128-99(2011) Standard Test Method for Maximum Pore Diameter and Permeability of Rigid Porous Filters for Laboratory UseAvailable at: http://www.astm.org/search/fullsite-search.html?query=maximum%20pore%20diameter%20and%20permeability%20of%20rigid%20porous%20filters%20
. Accessed January 30, 2016
17. Grulke EABrandrupt J, Immergut EH, Grulke ESolubility parameter valuesin Polymer Handbook2000New YorkJohn Wileypp. 519560