Implications of variability on medical device chemical equivalence assessment.

Chemical equivalence testing can be used to assess the biocompatibility implications of a materials or manufacturing change for a medical device. This testing can provide a relatively facile means to evaluate whether the change may result in additional or different toxicological concerns. However, one of the major challenges in the interpretation of chemical equivalence data is the lack established criteria for determining if two sets of extractables data are effectively equivalent. To address this gap, we propose a two-part approach based upon a relatively simple statistical model. First, the probability of a false positive conclusion, wherein there is an incorrectly perceived increase for a given analyte in the comparator relative to the baseline device, can be reduced to a prescribed level by establishing an appropriate acceptance criterion for the ratio of the observed means. Second, the probability of a false negative conclusion, where an actual increase in a given analyte cannot be discerned from the test results, can be minimized by specifying a limiting value of applicability based on the margin of safety (MoS) of the analyte. This approach provides a quantitative, statistically motivated method to interpret chemical equivalence data, despite the relatively high intrinsic variability and small number of replicates typically associated with a chemical characterization evaluation.

Modeling extraction of medical device polymers for biocompatibility evaluation.

Extraction testing is critical for biocompatibility evaluation of medical devices, whether to generate samples for biological testing or form the basis for toxicological risk assessment. However, it is not always clear how to compare extraction testing between different extraction conditions and sample geometries. We employ a physics-based model to elucidate the theoretical impact of extraction conditions, sample geometry and material properties on extraction efficiency (M/M0) and extract concentration (C/C0) for single-step and iterative/exhaustive extraction test methods. The model is specified by three parameters: thermodynamic contributions (Ψ), kinetic contributions (τ), and number of extraction iterations (N). We find that over the range of typical parameters for single-step extractions, M/M0 only approaches one (complete exhaustion) for relatively large values of Ψ (≥10) and τ(≥1). Further, the model suggests that test article geometry and solvent volume can have a dramatic and sometimes opposing effect on M/M0 and C/C0. Our results imply that iterative extractions can be approximated as a single-step extraction with scaled parameters Ψ' = ΨN and τ' = τN. The model provides a framework to reduce the biocompatibility evaluation test burden by optimizing test article and extraction condition selection and guiding development of new test protocols.

Predicting Solute Diffusivity in Polymers Using Time-Temperature Superposition.

We demonstrate a novel application of the time-temperature superposition (TTS) principle to predict solute diffusivity D in glassy polymers using atomistic molecular dynamics simulations. Our TTS approach incorporates the Debye-Waller factor ⟨u2⟩, a measure of solute caging, along with concepts from thermodynamic scaling methods, allowing us to balance contributions to the dynamics from temperature and ⟨u2⟩ using adjustable parameters. Our approach rescales the solute mean-squared displacement curves at several temperatures into a master curve that approximates the diffusive dynamics at a reference temperature, effectively extending the simulation time scale from nanoseconds to seconds and beyond. With a set of "universal" parameters, this TTS approach predicts D with reasonable accuracy in a broad range of polymer/solute systems. Using TTS greatly reduces the computational cost compared to standard MD simulations. Thus, our method offers a means to rapidly and routinely provide order-of-magnitude estimates of D using simulations.

Nitinol Release of Nickel under Physiological Conditions: Effects of Surface Oxide, pH, Hydrogen Peroxide, and Sodium Hypochlorite.

Nitinol is a nickel-titanium alloy widely used in medical devices for its unique pseudoelastic and shape-memory properties. However, nitinol can release potentially hazardous amounts of nickel, depending on surface manufacturing yielding different oxide thicknesses and compositions. Furthermore, nitinol medical devices can be implanted throughout the body and exposed to extremes in pH and reactive oxygen species (ROS), but few tools exist for evaluating nickel release under such physiological conditions. Even in cardiovascular applications, where nitinol medical devices are relatively common and the blood environment is well understood, there is a lack of information on how local inflammatory conditions after implantation might affect nickel ion release. For this study, nickel release from nitinol wires of different finishes was measured in pH conditions and at ROS concentrations selected to encompass and exceed literature reports of extracellular pH and ROS. Results showed increased nickel release at levels of pH and ROS reported to be physiological, with decreasing pH and increasing concentrations of hydrogen peroxide and NaOCl/HOCl having the greatest effects. The results support the importance of considering the implantation site when designing studies to predict nickel release from nitinol and underscore the value of understanding the chemical milieu at the device-tissue interface.

Relations Between Dynamic Localization and Solute Diffusion in Polymers.

Various public health concerns can arise from the unintended leaching of additives and impurities from polymeric medical devices or food packaging, which is directly related to each solute's diffusivity D. Both experimental and simulation methods can be used to quantify D, but slow diffusion at physiologic temperature in glassy polymers can render these approaches impractical. Here, we investigate a simulation approach with the potential to more rapidly calculate D. Specifically, we examine links between dynamic localization, characterized by the Debye-Waller factor, ⟨u2⟩, and D in a variety of polymer/solute systems using atomistic molecular dynamics (MD) simulations. Using short, high-temperature MD simulations to estimate D at physiologic temperature, we find that the relation ln D ∝ 1/⟨u2⟩ quantitatively predicts D for small solutes and produces an upper-bound estimate of D for larger solutes. Upper-bound estimates are useful in certain contexts, and we compare our results with another approach for determining upper bounds, the Piringer model, to show where each method may be useful. Then, we examine a modified relation where the Debye-Waller factor is rescaled by the mode coupling temperature Tc, which can produce better estimates of D if Tc is carefully chosen. Last, we compare our approach with several other models that relate temperature or localized dynamics with diffusivity. Although each of these approaches can be used to model D across wide temperature ranges using one or more adjustable parameters, none of them are truly predictive in glassy polymers. Further developments are needed to predict the optimal values of the adjustable parameters a priori.

Temperature dependence of nickel ion release from nitinol medical devices.

Nitinol exhibits unique (thermo)mechanical properties that make it central to the design of many medical devices. However, nitinol nominally contains 50 atomic percent nickel, which if released in sufficient quantities, can lead to adverse health effects. While nickel release from nitinol devices is typically characterized using in vitro immersion tests, these evaluations require lengthy time periods. We have explored elevated temperature as a potential method to expedite this testing. Nickel release was characterized in nitinol materials with surface oxide thickness ranging from 12 to 1564 nm at four different temperatures from 310 to 360 K. We found that for three of the materials with relatively thin oxide layers, ≤ 87 nm nickel release exhibited Arrhenius behavior over the entire temperature range with activation energies of 80 to 85 kJ/mol. Conversely, the fourth ''black-oxide'' material, with a much thicker, complex oxide layer, was not well characterized by an Arrhenius relationship. Power law release profiles were observed in all four materials; however, the exponent from the thin oxide materials was approximately 1/4 compared with 3/4 for the black-oxide material. To illustrate the potential benefit of using elevated temperature to abbreviate nickel release testing, we demonstrated that a > 50 day 310 K release profile could be accurately recovered by testing for less than 1 week at 340 K. However, because the materials explored in this study were limited, additional testing and mechanistic insight are needed to establish a protective temperature scaling that can be applied to all nitinol medical device components.

Leveraging Extraction Testing to Predict Patient Exposure to Polymeric Medical Device Leachables Using Physics-based Models.

Toxicological risk assessment approaches are increasingly being used in lieu of animal testing to address toxicological concerns associated with release of chemical constituents from polymeric medical device components. These approaches currently rely on in vitro extraction testing in aggressive environments to estimate patient exposure to these constituents, but the clinical relevance of the test results is often ambiguous. Physics-based mass transport models can provide a framework to interpret extraction test results to provide more clinically relevant exposure estimates. However, the models require system-specific material properties, such as diffusion (D) and partition coefficients (K), to be established a priori for the extraction conditions. Using systems comprised high-density polyethylene and 4 different additives, we demonstrate that these properties can be quantified through standard extraction testing in hexane and isopropyl alcohol. The values of D and K derived in this manner were consistent with theoretical predictions for these quantities. Based on these results, we discuss both the challenges and benefits to leveraging extraction data to parameterize physics-based exposure models. Our observations suggest that clinically relevant, yet still conservative, exposure dose estimates provided by applying this approach to a single extraction measurement can be more than 100 times lower than would be measured under typical aggressive extraction conditions. However, to apply the framework on a routine basis, limiting values of D and K must be established for device-relevant systems either through the aggregation and analysis of more extensive extraction test data and/or advancements in theoretical and computational modeling efforts to predict these quantities.

Strategies for Rapid Risk Assessment of Color Additives Used in Medical Devices.

Many polymeric medical devices contain color additives for differentiation or labeling. Although some additives can be toxic under certain conditions, the risk associated with the use of these additives in medical device applications is not well established, and evaluating their impact on device biocompatibility can be expensive and time consuming. Therefore, we have developed a framework to conduct screening-level risk assessments to aid in determining whether generating color additive exposure data and further risk evaluation are necessary. We first derive tolerable intake values that are protective for worst-case exposure to 8 commonly used color additives. Next, we establish a model to predict exposure limited only by the diffusive transport of the additive through the polymer matrix. The model is parameterized using a constitutive model for diffusion coefficient (D) as a function of molecular weight (Mw) of the color additive. After segmenting polymer matrices into 4 distinct categories, upper bounds on D(Mw) were determined based on available data for each category. The upper bounds and exposure predictions were validated independently to provide conservative estimates. Because both components (toxicity and exposure) are conservative, a ratio of tolerable intake to exposure in excess of one indicates acceptable risk. Application of this approach to typical colored polymeric materials used in medical devices suggests that additional color additive risk evaluation could be eliminated in a large percentage (≈90%) of scenarios.

Predicting Plasma Free Hemoglobin Levels in Patients Due to Medical Device-Related Hemolysis.

Blood passage through medical devices can cause hemolysis and increased levels of plasma free hemoglobin (pfH) that may lead to adverse effects such as vasoconstriction and renal tubule injury. Although the hemolytic potential of devices is typically characterized in vitro using animal blood, the results can be impacted by various blood parameters, such as donor species. Moreover, it is unclear how to relate measured in vitro hemolysis levels to clinical performance because pfH accumulation in vivo depends on both hemolysis rate and availability of plasma haptoglobin (Hpt) that can bind and safely eliminate pfH. To help to address these uncertainties, we developed a biokinetic model linking in vivo hemolysis rates to time-dependent pfH and Hpt concentrations. The model was initially parameterized using studies that characterized baseline levels and evolution of pfH and Hpt after introduction of excess pfH in humans. With the biokinetic parameters specified, the model was applied to predict hemolysis rates in three patient groups undergoing cardiopulmonary bypass surgery. The congruity of the model with these clinical data suggests that it can infer in vivo hemolysis rates and provide insight into pfH levels that may cause concern. The model was subsequently used to evaluate acceptance threshold hemolysis values proposed in the literature for chronic circulatory assist blood pumps and to assess the impact of patient weight on pfH accumulation using simple scaling arguments, which suggested that identical hemolysis index values may increase pfH levels nearly threefold in 10 kg pediatric patients compared with 80 kg adults.

Renal Toxicodynamic Effects of Extracellular Hemoglobin After Acute Exposure.

Plasma hemoglobin (Hb) is elevated in some hematologic disease states, during exposures to certain toxicants, and with the use of some medical devices. Exposure to free Hb can precipitate oxidative reactions within tissues and alter the normal physiological function of critical organ systems. As kidney structures can be highly sensitive to Hb exposures, we evaluated the acute dose dependent renal toxicologic response to purified Hb isolated from RBCs. Male Hartley guinea pigs (n = 5 per group) were dosed with 0.9% saline (2 ml), 15, 75, 150, or 300 mg of purified Hb, infused over a 2-h period. The primary endpoints of this study were to define toxicokinetic parameters after increasing doses of purified Hb, identify clinically recognized and experimental markers of acute kidney injury (AKI), and determine relevant toxicological parameters and potential causes of renal toxicity in this model. Experimental findings demonstrated a dose dependent increase in Cmax after a 2-h infusion, which correlated with an elevation in serum creatinine, renal Kim-1 mRNA expression and increased urinary Kim-1. Renal NGAL mRNA expression and urinary NGAL excretion were also increased in several groups, but these parameters did not correlate with exposure. Iron increased in the renal cortex as Hb exposure increased and its deposition colocalized with 4-hydroxy-nonenal and 8-Oxo-2-deoxyguanosine immune reactivity, suggesting oxidative stressors may contribute to AKI in animals exposed to Hb. The results presented here suggest that Cmax may effectively predict the risk of AKI in normal healthy animals exposed to Hb.

Predicting patient exposure to nickel released from cardiovascular devices using multi-scale modeling.

Many cardiovascular device alloys contain nickel, which if released in sufficient quantities, can lead to adverse health effects. However, in-vivo nickel release from implanted devices and subsequent biodistribution of nickel ions to local tissues and systemic circulation are not well understood. To address this uncertainty, we have developed a multi-scale (material, tissue, and system) biokinetic model. The model links nickel release from an implanted cardiovascular device to concentrations in peri-implant tissue, as well as in serum and urine, which can be readily monitored. The model was parameterized for a specific cardiovascular implant, nitinol septal occluders, using in-vitro nickel release test results, studies of ex-vivo uptake into heart tissue, and in-vivo and clinical measurements from the literature. Our results show that the model accurately predicts nickel concentrations in peri-implant tissue in an animal model and in serum and urine of septal occluder patients. The congruity of the model with these data suggests it may provide useful insight to establish nickel exposure limits and interpret biomonitoring data. Finally, we use the model to predict local and systemic nickel exposure due to passive release from nitinol devices produced using a wide range of manufacturing processes, as well as general relationships between release rate and exposure. These relationships suggest that peri-implant tissue and serum levels of nickel will remain below 5 µg/g and 10 µg/l, respectively, in patients who have received implanted nitinol cardiovascular devices provided the rate of nickel release per device surface area does not exceed 0.074 µg/(cm2 d) and is less than 32 µg/d in total. STATEMENT OF SIGNIFICANCE: The uncertainty in whether in-vitro tests used to evaluate metal ion release from medical products are representative of clinical environments is one of the largest roadblocks to establishing the associated patient risk. We have developed and validated a multi-scale biokinetic model linking nickel release from cardiovascular devices in-vivo to both peri-implant and systemic levels. By providing clinically relevant exposure estimates, the model vastly improves the evaluation of risk posed to patients by the nickel contained within these devices. Our model is the first to address the potential for local and systemic metal ion exposure due to a medical device and can serve as a basis for future efforts aimed at other metal ions and biomedical products.

Improving risk assessment of color additives in medical device polymers.

Many polymeric medical device materials contain color additives which could lead to adverse health effects. The potential health risk of color additives may be assessed by comparing the amount of color additive released over time to levels deemed to be safe based on available toxicity data. We propose a conservative model for exposure that requires only the diffusion coefficient of the additive in the polymer matrix, D, to be specified. The model is applied here using a model polymer (poly(ether-block-amide), PEBAX 2533) and color additive (quinizarin blue) system. Sorption experiments performed in an aqueous dispersion of quinizarin blue (QB) into neat PEBAX yielded a diffusivity D = 4.8 × 10-10 cm2  s-1 , and solubility S = 0.32 wt %. On the basis of these measurements, we validated the model by comparing predictions to the leaching profile of QB from a PEBAX matrix into physiologically representative media. Toxicity data are not available to estimate a safe level of exposure to QB, as a result, we used a Threshold of Toxicological Concern (TTC) value for QB of 90 µg/adult/day. Because only 30% of the QB is released in the first day of leaching for our film thickness and calculated D, we demonstrate that a device may contain significantly more color additive than the TTC value without giving rise to a toxicological concern. The findings suggest that an initial screening-level risk assessment of color additives and other potentially toxic compounds found in device polymers can be improved. © 2017 Wiley Periodicals, Inc. J Biomed Mater Res Part B: Appl Biomater, 106B: 310-319, 2018.

Characterizing the free volume of ultrahigh molecular weight polyethylene to predict diffusion coefficients in orthopedic liners.

Liners used in orthopedic devices are often made from ultrahigh molecular weight polyethylene (UHMWPE). A general predictive capability for transport coefficients of small molecules in UHMWPE does not exist, making it difficult to assess properties associated with leaching or uptake of small molecules. To address this gap, we describe here how a form of the Vrentas-Duda free volume model can be used to predict upper-bound diffusion coefficients (D) of arbitrary molecules within UHMWPE on the basis of their size and shape. Within this framework, the free-volume microstructure of UHMWPE is defined by analysis of a curated set of model diffusants. We determined an upper limit on D for vitamin E, a common antioxidant added to UHMWPE, to be 7.1 × 10-12 cm2  s-1 . This means that a liner that contains 0.1 wt % or less Vitamin E and has <120 cm2 patient contacting surface area would elute <100 µg/day of vitamin E. Additionally, the model predicts that squalene and cholesterol-two pro-oxidizing biological compounds-do not penetrate over 820 µm into UHMWPE liners over the course of 5 years because their D is ≤7.1 × 10-12 cm2  s-1 . © 2017 Wiley Periodicals, Inc. J Biomed Mater Res Part B: Appl Biomater, 106B: 2393-2402, 2018.

Conservative Exposure Predictions for Rapid Risk Assessment of Phase-Separated Additives in Medical Device Polymers.

A novel approach for rapid risk assessment of targeted leachables in medical device polymers is proposed and validated. Risk evaluation involves understanding the potential of these additives to migrate out of the polymer, and comparing their exposure to a toxicological threshold value. In this study, we propose that a simple diffusive transport model can be used to provide conservative exposure estimates for phase separated color additives in device polymers. This model has been illustrated using a representative phthalocyanine color additive (manganese phthalocyanine, MnPC) and polymer (PEBAX 2533) system. Sorption experiments of MnPC into PEBAX were conducted in order to experimentally determine the diffusion coefficient, D = (1.6 ± 0.5) × 10-11 cm2/s, and matrix solubility limit, C s = 0.089 wt.%, and model predicted exposure values were validated by extraction experiments. Exposure values for the color additive were compared to a toxicological threshold for a sample risk assessment. Results from this study indicate that a diffusion model-based approach to predict exposure has considerable potential for use as a rapid, screening-level tool to assess the risk of color additives and other small molecule additives in medical device polymers.

Solvent or thermal extraction of ethylene oxide from polymeric materials: Medical device considerations.

Ethylene oxide (EO) gas is commonly used to sterilize medical devices. Bioavailable residual EO, however, presents a significant toxicity risk to patients. Residual EO is assessed using international standards describing extraction conditions for different medical device applications. We examine a series of polymers and explore different extraction conditions to determine residual EO. Materials were sterilized with EO and exhaustively extracted in water, in one of three organic solvents, or in air using thermal desorption. The EO exhaustively extracted varies significantly and is dictated by two factors: the EO that permeates the material during sterilization; and the effectiveness of the extraction protocol in flushing residual EO from the material. Extracted EO is maximized by a close matches between Hildebrand solubility parameters δpolymer , δEO , and δsolvent . There remain complexities to resolve, however, because maximized EO uptake and detection are accompanied by great variability. These observations may inform protocols for material selection, sterilization, and EO extraction. © 2017 Wiley Periodicals, Inc. J Biomed Mater Res Part B: Appl Biomater, 106B: 2455-2463, 2018.

Polymer degradation and drug delivery in PLGA-based drug-polymer applications: A review of experiments and theories.

Poly (lactic-co-glycolic acid) (PLGA) copolymers have been broadly used in controlled drug release applications. Because these polymers are biodegradable, they provide an attractive option for drug delivery vehicles. There are a variety of material, processing, and physiological factors that impact the degradation rates of PLGA polymers and concurrent drug release kinetics. This work is intended to provide a comprehensive and collective review of the physicochemical and physiological factors that dictate the degradation behavior of PLGA polymers and drug release from contemporary PLGA-based drug-polymer products. In conjunction with the existing experimental results, analytical and numerical theories developed to predict drug release from PLGA-based polymers are summarized and correlated with the experimental observations. © 2016 Wiley Periodicals, Inc. J Biomed Mater Res Part B: Appl Biomater, 105B: 1692-1716, 2017.

Controlled initial surge despite high drug fraction and high solubility.

Potential connections between release profiles and solvent evaporation rates alongside polymer chemistry were elucidated for the release of tetracycline hydrochloride from two different poly (d, l-lactide-co-glycolide) (PLGA) film matrices containing high drug fractions (50%, 30%, and 15%), and prepared at two distinct solvent evaporation rates. At highest tetracycline concentrations (50%), (i) the early release rates were ≤0.5 µg/min in all cases; (ii) release was linear from systems fabricated with lower lactic content and slower solvent evaporation rate and bimodal from systems fabricated with higher lactic content and faster evaporation rate; (iii) surface fractions covered by the drug were similar at both evaporation rates for 85:15 PLGA but very different for 50:50 PLGA, leading to unexpectedly reduced early release from 50:50 PLGA than from 85:15 PLGA when both the matrices were fabricated using a slower evaporation rate. These features remained unaffected in case of low drug concentration. Results suggested that during the formation of the drug-polymer microstructure, the combined effect of polymer chemistry and solvent evaporation rate sets apart the surface characteristics and the initial release profiles of systems containing high drug fraction, and an appropriate combination of these parameters may be utilized to control the early stage of drug release.

Communication: Relationship between solute localization and diffusion in a dynamically constrained polymer system.

We investigate the link between dynamic localization, characterized by the Debye-Waller factor, 〈u(2)〉, and solute self-diffusivity, D, in a polymer system using atomistic molecular dynamics simulations and vapor sorption experiments. We find a linear relationship between lnD and 1/〈u(2)〉 over more than four decades of D, encompassing most of the glass formation regime. The observed linearity is consistent with the Langevin dynamics in a periodically varying potential field and may offer a means to rapidly assess diffusion based on the characterization of dynamic localization.

A biokinetic model for nickel released from cardiovascular devices.

Many alloys used in cardiovascular device applications contain high levels of nickel, which if released in sufficient quantities, can lead to adverse health effects. While nickel release from these devices is typically characterized through the use of in-vitro immersion tests, it is unclear if the rate at which nickel is released from a device during in-vitro testing is representative of the release rate following implantation in the body. To address this uncertainty, we have developed a novel biokinetic model that combines a traditional toxicokinetic compartment model with a physics-based model to estimate nickel release from an implanted device. This model links the rate of in-vitro nickel release from a cardiovascular device to serum nickel concentrations, an easily measured endpoint, to estimate the rate and extent of in-vivo nickel release from an implanted device. The model was initially parameterized using data in the literature on in-vitro nickel release from a nickel-containing alloy (nitinol) and baseline serum nickel levels in humans. The results of this first step were then used to validate specific components of the model. The remaining unknown quantities were fit using serum values reported in patients following implantation with nitinol atrial occluder devices. The model is not only consistent with levels of nickel in serum and urine of patients following treatment with the atrial occluders, but also the optimized parameters in the model were all physiologically plausible. The congruity of the model with available data suggests that it can provide a framework to interpret nickel biomonitoring data and use data from in-vitro nickel immersion tests to estimate in-vivo nickel release from implanted cardiovascular devices.

Diffuse Interface Methods for Modeling Drug-Eluting Stent Coatings.

An overview of diffuse interface models specific to drug-eluting stent coatings is presented. Microscale heterogeneities, both in the coating and use environment, dictate the performance of these coatings. Using diffuse interface methods, these heterogeneities can be explicitly incorporated into the model equations with relative ease. This enables one to predict the complex microstructures that evolve during coating fabrication and subsequent impact on drug release. Examples are provided that illustrate the wide range of phenomena that can be addressed with diffuse interface models including: crystallization, constrained phase separation, hydrolytic degradation, and heterogeneous binding. Challenges associated with the lack of material property data and numerical solution of the model equations are also highlighted. Finally, in light of these potential drawbacks, the potential to utilize diffuse interface models to help guide product and process development is discussed.