Accelerated lifetime testing A research strategy that allows prediction of the performance of products over a long term based on tests conducted rapidly in the laboratory.
Accuracy The tendency of a test method to provide a result that is close to the correct result or the tendency of a statistical model to make a prediction that is close to a later observation.
Clinical relevance Conducting research in a way that closely mimics the conditions present in the clinic and in the patient.
Clinical significance A difference or change in product performance that actually affects the patient’s quality of life.
Design of experiments A research strategy that allows efficient screening and optimization of design factors.
Effective volume The volume of a hypothetical test specimen with uniform stress distribution, such as in pure tension, that would have the same probability of fracture as an actual specimen of interest with nonuniform stress, such as a specimen tested in flexure mode.
Finite-element analysis An engineering analysis method for predicting the development of stress and strain in structures or the flow of energy or fluids through those structures.
In vitro research The testing of materials, drugs, and devices that is conducted in a basic science laboratory instead of being conducted in living test subjects.
Precision The tendency of a test method to consistently provide the same result.
Specification A document that describes one or more mandatory test methods and, in some cases, minimum acceptable levels of performance for a product.
Statistical significance A difference or change in product performance that is unlikely to be caused by random sampling error.
Technique sensitivity The tendency of some products to perform differently when manipulated by different operators.
In vitro research is the testing of materials, drugs, and devices that is conducted in a basic science laboratory instead of being conducted in living test subjects. This could mean measuring the degradation of materials when they are exposed in a test chamber that simulates the chemical composition and temperature of the oral environment, or testing could include measuring the mechanical response in a load frame or passing ultrasonic pulses through test specimens. In other cases, visible light may be measured as light is reflected and transmitted by a specimen to determine the color shade and translucency of the material. Reflection of infrared radiation and x-rays may also be measured to determine the molecular structure and the crystal structure of materials, respectively. In vitro tests may be conducted using synthetic materials to hold the specimen in place, or the specimen may be held by cadaver tissues. Virtual testing may be performed using three-dimensional (3-D) computer models to predict the performance of dental prostheses made from different materials. These are only a few examples of the many in vitro test methods used in the research of dental materials and devices. Some tests involve placing materials in a culture of living human or bacterial cells and observing the cellular response. The interpretation of a living indicator provides a greater variety of factors to consider, so cell culture tests are discussed separately in Chapter 17, Biocompatibility .
The Role of In Vitro Research
Students and professionals who are familiar with the topic of evidence-based medicine may recall that in vitro research is often placed low in the hierarchy of evidence. The vertical location of each layer in the evidence pyramid ( Figure 18-1 ) indicates the relative weight that should be applied to that type of research in deciding whether a novel dental material or device performs well clinically. The width of each layer in the pyramid indicates the relative amount of that type of evidence that is available in the scientific literature. One important aspect of in vitro research is that there is a large volume of evidence available. However, each bit of in vitro evidence should be weighted lightly while making decisions about new dental products because those data—although intended to predict clinical performance—are not actually gathered in a clinical situation and may have little predictive power individually. Despite these drawbacks, in vitro research still has an important impact early in the process of developing a dental product. In vitro results can bolster confidence in the safety of a product (as measured in terms of chemical durability, mechanical durability, and lack of toxicity) and in the efficacy of a product (as measured in terms of esthetics, dimensional accuracy, or antimicrobial activity, for example). This confidence is necessary before conducting clinical studies to avoid endangering human test subjects and is necessary before conducting animal studies to avoid being wasteful in the allocation of those expensive and time-consuming studies.
How can we gain confidence in the safety of a product prior to conducting clinical trials?
Design of Experiments
There are many combinations of manufacturing parameters to be tested in the development of a new dental product. Besides the safety considerations discussed previously, the work requires a great number of specimens to be fabricated and tested. This means that it is important for in vitro tests to be precise, rapid, and inexpensive relative to clinical testing. Figure 18-2 shows a flowchart illustrating how the workflow might be conducted when optimizing the formulation of a dental composite that includes a novel comonomer. Optimization of a new dental composite is a topic that has been the focus of much previous research, so for researchers developing the flowchart, guessing as to which few design parameters will be the most important to control and vary is fairly easy. Other types of projects contain a greater number of variables. For example, dental implants sometimes have 25 different design features, and it may not be obvious from the outset which design features have the greatest impact on the performance of the implant. Even if each design parameter was tested in only two conditions, then a full factorial study would require fabricating and testing 2 25 = 33,554,432 groups of implants! In such cases, researchers employ powerful statistical methods with the assistance of research software (design of experiments [DOE] and accelerated lifetime testing [ALT]) or use engineering simulation software (finite-element analysis [FEA]) to conduct in vitro research as efficiently as possible.
DOE is a research strategy that involves three steps: (1) screening design factors to determine which ones have the greater effect on performance, (2) identifying the combination of choices for the important factors that correspond to maximum possible performance (the optimal formulation; i.e., best balance of properties among those being investigated), and (3) identifying the combination of choices near the optimal formulation that results in the lowest variability in performance (the least technique-sensitive formulation). The first step in DOE is usually conducted using orthogonal arrays. Orthogonal arrays are lists of various combinations of choices for the design factors. Each combination corresponds to a different formulation or test group, and the list is constructed to be as short as possible by making the simplifying assumption that the importance of each design factor is independent of the choices for the other factors (no interactive effects). Orthogonal arrays can be quite efficient. For example, only eight test groups are needed to screen seven factors when the factors have two settings each. The second step in DOE involves using response-surface methods or sequential minimum-energy design to efficiently explore several possible choices or levels for each remaining factor. After the design space has been explored, the design corresponding to the maximum performance (strength, patient satisfaction, shelf life, etc . ) or the minimum cost is identified. Compromise formulations can also be identified to satisfy multiple performance criteria simultaneously. The third step in DOE uses Taguchi robust design. This is similar to the second step, except minimum variability is treated as a performance objective, and a compromise formulation can be identified between minimizing variability and maximizing other performance measures. Minimizing variability is important for dental products because the dentists or dental lab technicians who will use the products have different levels of experience and have different personal habits of preparing teeth and manipulating materials. A technique-insensitive material system performs similarly regardless of the level of experience of the operator. In the case of a dental restorative material, for example, it is desirable for the product to have decreased technique sensitivity in addition to lending itself to forming restorations with a long average lifetime.
Why is it important for in vitro tests to be precise, rapid, and inexpensive compared with clinical testing?
Accelerated Lifetime Testing
ALT is used to predict the performance of dental materials over a long period in the patient’s mouth based on tests conducted rapidly in the research laboratory. This is accomplished by testing material specimens at several stress levels (all of them being greater than the stress level of the clinical case). A statistical model is fit to the data to forecast the lifetime that would probably be observed at clinical stress levels ( Figure 18-3 ). In ALT, the term stress is used in a general sense of the response of materials to an external stimulus and does not necessarily imply units of mechanical stress (such as MPa). Nonetheless, mechanical loading is the most common type of stressor used in ALT, but other challenges (i.e., temperature, humidity, etc.) may also be applied either singly or simultaneously.
ALT can be conducted as constant-stress accelerated lifetime testing (CSALT) or as step-stress accelerated lifetime testing (SSALT). In CSALT, each specimen encounters only one level of stress from the beginning of the test up to the moment when failure is recorded, such as one amplitude of stress in the case of a chewing simulation. In SSALT, a test specimen starts at a specified low stress. If the test unit does not fail at a specified time, stress on the unit is raised and held for a specified time, again. Stress is repeatedly increased until the test unit fails or the censoring time is reached. This leads to more rapid testing. However, researchers may have less confidence in clinical lifetimes forecasted from SSALT data than those forecasted from CSALT data.
It is important to remember that both types of predictions are extrapolations, and they include a great amount of uncertainty compared with clinical data. There is also a possibility that the levels of stressors, such as temperature or rate of mastication, may be elevated so greatly during ALT that degradation processes that would never occur in the oral cavity may be activated in addition to the processes that are being accelerated. In this case, inaccurate lifetime predictions are made. ALT can be used in conjunction with DOE by selecting clinical lifetime predictions as a performance measure to be maximized. In addition to predicting useful material lifetime in the oral cavity, ALT can also be used to predict the shelf-life of auxiliary dental materials such as alginate impression materials.
FEA is an engineering analysis method for predicting the development of mechanical stress and strain or predicting the flow of energy (heat, electricity, magnetic fields, etc.) or the flow of fluids through structures. Current applications of FEA are aided by 3-D models and specialized computer software. A solid model may be fabricated using computer-aided drafting (CAD) software, or the model may be imported from a clinical scanner (microcomputed tomography [micro-CT]; Figure 18-4, A ). This allows the modeling of precisely machined parts, such as dental implant components, and the modeling of parts with custom geometry, such as fixed dental prostheses and bones and other tissues (see Chapter 15, Digital Technology in Dentistry ).
The solid model is divided into thousands or millions of minuscule volume elements ( Figure 18-4, A ), and each element is assigned properties according to the material that constitutes that portion of the model. Using a greater number and smaller size of elements results in more accurate predictions but also requires more computing time and/or power. An important test of accuracy that is often neglected involves solving an FEA problem multiple times, with an increasing number of elements in each iteration. The results should converge on a limiting value. If such a convergence test is not present with the reported results, then the FEA predictions may not be accurate. The absolute level of stress predicted cannot be trusted in this case, but the pattern of the stress distribution still provides useful information regarding the locations of higher and lower stress concentration in a dental prosthesis ( Figure 18-4, B ).
The most common results to be predicted using FEA are the stress and strain resulting from mechanical loading. That type of analysis requires inputting material properties of Young’s modulus and Poisson’s ratio ( Chapter 4, Stress-Strain Properties ). The accuracy of material properties depends on the method used to measure them, and if inaccurate material properties are input to the model or if invalid assumptions are made regarding the anisotropy (direction dependence) of these properties, then the FEA results will be invalid. Stress and strain results can be exported to additional postprocessing software that combines these results with material fatigue testing data to predict the probability of failure by a specified elapsed time for each element in the finite-element model and to predict the probability of failure for a dental prosthesis or implant as a whole. FEA can be used in conjunction with DOE by selecting the probability of failure as a performance measure to be minimized.
What factors lead to a lack of accurate forecasting in ALT?
Clinical Relevance of In Vitro Tests
For some classes of dental materials, there are in vitro tests that can predict the performance of a given formulation or the comparative ranking for different formulations, but performance can be difficult to predict for other applications or classes of materials. For example, David Mahler showed that marginal fracture or “ditching” in dental amalgam fillings can be predicted by tests of creep resistance. His finding was that the more ditching-resistant amalgams lacked a γ 2 phase in the microstructure, and that the formation of the γ 2 phase could be avoided by increasing the copper content of the amalgam alloy. The resulting high-copper amalgams produced dental fillings with greatly improved longevity. On the other hand, Jack Ferracane compiled material property data from a variety of in vitro tests, and he concluded that the clinical failure of composite restorations is a complex, multifactorial process that cannot be predicted from an in vitro test or a combination of tests. However, in the early development and screening stages of new dental materials, in vitro tests remain the only tools we have and thus the best tools that we can use to make an educated guess regarding which material formulations will exhibit superior performance.
Precision Versus Clinical Relevance
There are often several competing in vitro tests that can be used to measure a single material property. For example, dentin bond strength can be measured by a microtensile bond-strength test, a button shear test, or a crown-removal test ( Figure 18-5 ). There are even several variations in the geometry of the test specimen or the method of load application to choose from when conducting each of these three test methods. In these cases, there is usually not a single test method that has superior performance in testing all kinds of hypotheses related to a particular material property. Rather, the various test methods are usually distributed across a spectrum that has a high level of clinical relevance on one end and a high level of precision on the other end.