An MD&DI March 1998 Column
How to Meet the New Simulation Testing Requirements
In addition to fulfilling the design validation requirements of the new GMPs, a properly constructed simulation test provides a convenient, economical, and reproducible means of understanding how devices will actually perform.
Michael DruesUnlike traditional mechanical or biological testing, simulation testing seeks to predict how a device will actually function in or around a human body. The new FDA quality system regulation (QSR) requires testing under simulated or actual use conditions for all nonexempt Class II and Class III medical devices. Various simulation options are available to meet this requirement, each with its own advantages and drawbacks. This article examines the rationale behind simulation testing and presents a 10-step approach to developing an effective simulation strategy. For the purpose of illustration, the discussion focuses on simulation testing of a balloon angioplasty catheter; however, the general ideas presented are applicable to virtually any medical device.
During the development of any new medical device, many types of tests are routinely conducted, such as electrical, physical, chemical, and performance. Product or development tests can be divided into the general categories of mechanical and biological.
Mechanical testing essentially checks the design of the device, evaluating lengths, diameters, wall thicknesses, durometers, moduli, and similar mechanical and physical properties. A commonly used mechanical test for balloon catheters, for example, is the so-called burst test, which entails inflating the angioplasty balloon until it bursts. The pressure at which the balloon fails is then recorded and used to determine the maximum recommended inflation pressure for the product. While burst pressure is an important engineering parameter that helps to establish the safety limits of the device, it is generally determined without reference to the in vivo environment in which the device will perform. True, most balloon burst tests are performed in a bath of water at body temperature because heat will affect the mechanical properties of the balloon materials, but with the exception of temperature in this example, mechanical testing is mostly unrelated to the anatomy and physiology of the human body.
Biological testing analyzes the interaction of the device with human cells and tissues. Geometric and physical design parameters are usually of little significance. Examples of biological tests include biocompatibility, pyrogenicity, and mutagenicity assays. In regard to the angioplasty catheter, biological testing would be used to ensure that the surface of the catheter has a low thrombogenicity to minimize the formation of blood clots during use. Like mechanical testing, biological testing is, for the most part, performed without regard to the actual in vivo environment.
In contrast to mechanical or biological testing, simulation testing focuses on the actual in
|Type of Test||Area of Concentration|
|Mechanical testing||Design of device|
|Biological testing||Materials in device|
|Simulation testing||Environment around device|
Table I. Comparison of testing options and the areas of focus for each one.
A given simulation may not be limited to a particular device or product. Rather, it could be used for any device that performs in the same environment and addresses the same diseases or conditions. For example, a simulation developed for angioplasty catheters may be equally suited for stents, atherectomy devices, and minimally invasive bypass products.
GMP SIMULATION TESTING REQUIREMENTS
The design control section of the QSR requires manufacturers to establish and maintain procedures for validating device design. The document states that design validation should be performed under defined operating conditions on initial production units, lots, batches, or their equivalents. Design validation should ensure that devices conform to defined user needs and intended uses and should include testing of production units under actual or simulated use conditions. The results of the design validation, including identification of its design, methods, date, and individuals performing the testing, should be documented in the design history file.1
The operative phrase here is testing under actual or simulated use conditions. Thus, the regulation requires conditions-of-use testing, but it is up to the manufacturer to determine how to meet this requirement. Because this regulation only recently took effect, no precedents currently exist, and it will be some time before information regarding implementation techniques is available through the Freedom of Information Act.
APPROACHES TO SIMULATION TESTING
Four general approaches to simulation testing involve physical models; theoretical and computational models; animal models; and clinical, or human, models, respectively (Table II).
|In vitro||In vitro||In vivo||In vivo|
|Mechanical model||Theoretical/ |
|Nonhuman animal model||Clinical, or human, model|
Table II. The four general approaches to simulation testing including the environment and models involved.
Physical, or benchtop, simulations seek to mimic the anatomy and physiology of the human body in vitro. Models are typically made of synthetic materials and may be designed to have the same look and feel as a patient. They are usually more convenient and easier to use than in vivo simulations because they can sit in the lab until they are needed. In vitro simulation methods usually offer a higher degree of reproducibility and less variability than in vivo methods.
Theoretical simulation techniques include finite element analysis and computational fluid dynamics. They are usually limited by the assumptions made in their derivation and the computational power of the computer. Unfortunately, it may be difficult to quantitatively describe a device in sufficient detail to yield realistic resultsin other words, the designer can't introduce a catheter into the computation model by inserting it into the floppy drive of the computer!
Animal models include live canine, porcine, or bovine specimens, among others. While these animals do offer an in vivo environment, their anatomy and physiology may differ significantly from that of a human. They can also be expensive and require specialized facilities, limiting their in-house use. Reproducibility may be an issue because both inter- and intrasubject variability are difficult to control.
Human models present a true in vivo environment, but clinical testing is usually conducted near the end of the product development cycle and may not be appropriate for prototype or early developmental purposes. Clinical testing is subject to its own set of regulations (good clinical practices) and requires prior approval from an institutional review board. As with animal models, human testing is expensive and time-consuming, and even after careful patient selection both inter- and intrasubject variability are difficult to control.
There is no perfect simulation solution, but depending on the device, there are advantages and disadvantages to each simulation testing option.
A 10-STEP APPROACH
To develop an effective and reproducible simulation, the manufacturer must begin by asking the right questions. To simplify this process, here are 10 questions that will help guide decision makers through the process.
Is a Simulation Necessary? There are many reasons for conducting a simulation. Before deciding which strategy is best, the manufacturer must determine why a simulation is needed in the first place. Some common justifications for simulations include meeting the new GMP requirements, developing an R&D tool for prototype testing and design comparisons, predicting or recreating product failures, developing a training aid for clinicians, and demonstrating the use of a new product at clinical trade shows. A simulation could serve more than one purpose; however, different objectives might require different strategies, and more than one type of simulation might be needed. The manufacturer should identify the most important objective and develop a simulation strategy to meet it. To meet additional objectives, the designer might have to modify the simulation or create a new one.
Biosimulation tools recreate the look, feel, and response of the human body. Illustration courtesy of Vascular Sciences (Natick, MA).
For example, an R&D simulation model for testing a balloon catheter might be constructed of transparent materials to permit visual observation of device performance. In contrast, a simulation model for physician training might be colored to illustrate patient anatomy better and to prevent the physician from seeing inside the patient.
Which Simulation Option is Most Appropriate? After identifying the objective of the simulation, the manufacturer must decide which simulation strategy will best meet that objective. From the four main simulation strategies (Table II), designers must choose the most appropriate one for the application, keeping in mind that there is no perfect solution. For evaluating a balloon catheter prototype, a physical (in vitro) simulation approach might prove most appropriate.
What Will the Simulation Cover? Before building a physical simulation model, the manufacturer must define what exactly will be simulated. For example, should the model simulate a normal situation or an abnormal condition, perhaps resulting from illness or injury? In the case of the angioplasty catheter, the model should simulate the normal in vivo environment, except within the area to be treated. In other words, only a portion of the model will contain any pathology (i.e., a stenosis) and the remainder will simulate healthy tissue.
What Is the Structural Environment? The structural environment refers to the anatomy (normal structure) and pathology (abnormal structure) to be simulated. Which organs and tissues are involved? What are their mechanical properties? What are their relative shapes and sizes? How are they connected or attached? To simulate a portion of the cardiovascular system, the designer must determine the geometries and dimensions of the relevant blood vessels, keeping in mind that a high degree of anatomical variability will be observed in vivo. Also, the mechanical properties of the blood vessels and surrounding tissues must be determined, and any changes in these properties caused by pathology should be well represented. For example, arteries usually become less compliant when atherosclerotic material is present; therefore, different tubing would be used to simulate normal and diseased vessels.
What's the Functional Environment? The functional environment encompasses both physiology (normal function) and pathophysiology (abnormal function). Here, the designer must consider how the organs and tissues work together and what sort of movement or flow might be encountered. For example, stenotic lesions in arteries cause an increase in blood flow and a decrease in local blood pressure. Following angioplasty, the lumenal diameter returns to normal and the pressure drop across the lesion falls to zero. This complex relationship can be directly observed using a properly designed physical simulation.
Which Parameters Might Influence the Function of the Device? Here, the idea is to identify all possible parameters in the in vivo environment that might influence how the device functions. Prioritizing these factors will come laterfor now, the designer should list everything that could possibly influence the function of the product.
It might help to break this question into two parts. First, the designer might consider which mechanical or physical parameterssuch as blood pressure, blood flow, temperature, and viscositycould influence the function of the device. Second, the designer might ask which material or tissue properties would come into play. For instance, is the elasticity or compliance of the tissue important? What about friction between tissues or between tissues and the product? All possible factors that could conceivably influence the function of the product must be identified.
Which Parameters Are Most Important? After generating a comprehensive list of the parameters that could affect device function, the designer should prioritize them. The goal is to determine which parameters will have the greatest influence on how the product performs and to optimize the simulation accordingly. Of course, it is impossible to optimize all of the parameters simultaneously in one simulation, and manufacturers might have to create different simulations to optimize each group of parameters separately.
Can Off-the-Shelf Components Be Used in the Simulation? It is usually quicker and cheaper to use off-the-shelf rather than custom products wherever possible. For example, if blood flow is important in the simulation, designers must consider whether an off-the-shelf pump would suffice, or whether a custom pumping system would have to be built. Can the flow be continuous, or must it be pulsatile? Will it be necessary to simulate changes in heart rate, cardiac output, or systolic/diastolic ratio? In general, the more precise the simulation requirements, the more difficult it will be to identify suitable off-the-shelf components. One strategy might be to start with a simple simulation and add more capabilities incrementally.
How Should the Simulation Be Designed? The actual simulation can now be designed. At this point, the designer should study the responses to the previous questions. If using off-the-shelf components, he or she should review the instructions and any other product information that may have been included. Since the design of each simulation depends on the responses to the previous questions, some general suggestions regarding material selection and fabrication techniques are provided below.
Material Selection. The choice of materials used in a simulation will to a large extent determine how realistically the model simulates the in
Because simulations do not come in contact with patients, the entire universe of materials is availablenot simply the so-called biocompatible materials. Even materials not typically used in the medical industry can be used in a simulation. The key is to match the mechanical properties of the simulation material as closely as possible to the mechanical properties of the native tissue. Of course, diseased tissue may have significantly different mechanical properties than normal tissue, and these differences must be factored into the model design as well. Consequently, a number of very different materials might be used in each simulation, including rubbers such as latex and silicone, and polymers such as polyurethane and polyethylene. By carefully controlling such parameters as wall thickness and durometer, a good match in mechanical properties can be achieved.
Design and Fabrication. After appropriate materials have been selected, the next step is to fabricate them into the tissue geometry. Obtaining realistic geometries can be difficult, though obviously the more precisely the simulated geometry matches that of the native tissue, the more realistic the simulation.
The geometric data needed for fabrication can be obtained in two ways. The traditional approach is to obtain data from the literature on morphology or from cadaver measurements. While not a bad approximation, this method is time-consuming and permits a large degree of error. A better method would be to get the geometric data directly from a patient or from sources such as the Visible Human Project.2
After collecting the appropriate geometric data, the manufacturer must fabricate the simulation model. Depending on the complexity of the part and the type of materials used, the model might be molded or extruded. For complex geometries, however, molding techniques may become cumbersome and expensive. Rapid prototyping techniques such as stereolithography offer a relatively inexpensive alternative. Techniques are being developed that allow actual patient data (obtained from MRI or spiral-CT images) to be fed directly into the rapid prototyping system, thus replicating the patient's anatomy exactly. This technology allows for the production of extremely realistic simulations.
How Is the Simulation Validated? Once the simulation model has been constructed, it is almost ready for use. However, one last critical step remains: validation.
Validation is important for three reasons. First, it will help determine the degree of realism of the simulationin other words, how accurately the simulation performs compared to the in vivo environment. If the intent is to simulate blood flow in the carotid artery, for example, validation will reveal how flow rate in the simulation compares to the flow rate in the patient.
Validation also identifies the limitations of the simulation. Especially in theoretical simulations, it is important to understand not only the accuracy of the model but also the circumstances under which it breaks down. A simulation might be highly accurate under normal situations, but if an abnormality is introduced, the simulated results might not be representative of a similar abnormality in vivo.
Finally, if data from the simulation will be used in support of an FDA submission, the simulation must be validated. The stronger the validation, the stronger the submission.
In general, two validation strategies are available: quantitative and qualitative. The first involves collecting quantifiable data from the simulation and comparing them to data collected in vivo under similar conditions. Collecting quantitative data usually involves the use of instrumentationfor example, pressure transducers to record simulated blood pressure. Common diagnostic technologies such as ultrasound or MRI might also be used. Of course, technologies such as ultrasound are designed for use on biological tissue, and since the simulation will probably be made of inert materials, some modifications may be necessary. (Such modifications are certainly feasible, but the details are beyond the scope of this article.)
In qualitative validation, experienced users (usually clinicians) use the device in the simulation following the same protocols they would follow when using the product in a patient. After the simulated procedure, users convey in as much detail as possible how the device performed in comparison to their experience of using a comparable device in a patient. To test a new product, the manufacturer should identify current users of similar products. For obvious reasons, users having the most clinical experience will typically yield the best results. When used properly, the "touchy-feely" data generated in the qualitative validation are as important as the numerical data obtained from the quantitative validation.
A strong validation strategy involves a combination of both quantitative and qualitative validations. Conducting only one type of validation leaves open the possibility that some important piece of information has been missed. When used together, the results help maximize the accuracy and degree of realism of the simulation.
Interest in building realistic simulations has recently increased, partly in response to the new GMP requirements; however, manufacturers will find other advantages in developing simulations, too. By allowing designers to evaluate early prototype designs realistically in the R&D lab, simulation testing can help decrease product development time. Simulation testing also helps to minimize expensive and controversial animal testing while furnishing a better understanding of how devices performall in a cost-effective, convenient, and reproducible fashion.
As medical devices become increasingly more complex, the need for robust simulation tools beyond simple anatomical models will grow more pressing. Although creating realistic simulations can seem overwhelming, manufacturers can, with proper consideration and planning, design an effective strategy to meet specific needs and objectives.
1. "Design Control Guidance for Medical Device Manufacturers," Rockville, MD, FDA, Center for Devices and Radiological Health, p 33, March 11, 1997.
2. For additional information on the Visible Human Project, see www.nlm.nih.gov/research/visible/visible_human.html.
Michael Drues, PhD, is president and chief technical officer for Vascular Sciences in Natick, MA, and an adjunct professor of medicine at Northeastern University in Boston.