Computational Modeling and Simulation Expanding in Orthopaedic Device Life Cycles

CMS 1

Computational modeling and simulation (CM&S) has the potential to revolutionize medical devices by accelerating innovation and providing comprehensive evidence of long-term safety. CM&S accomplishes these goals by providing a flexible and open platform to determine performance benchmarks, assess design parameter interdependencies, evaluate a variety of use conditions and visualize complex processes, and one day becoming a core element of device submissions and approvals. But the credibility of a computational model must be established before using model results for decision-making. To this end, FDA’s Center for Devices and Radiological Health (CDRH) and the American Society of Mechanical Engineers (ASME) are drafting verification and validation (V&V) guidelines for CM&S.

These guidelines will enable modeling to be a credible and common means for device companies and FDA to demonstrate the safety of medical devices, and thereby ensure safety, reduce cost and accelerate the pathway toward “first in the world” access to the newest and most innovative medical products.

CM&S in the Medical Device Life Cycle 

There are several phases in the life cycle of an orthopaedic device wherein CM&S plays a prominent role. These include conceptualization, design V&V, marketing claims and postmarket evaluation. Morphological and statistical shape analyses can be used at the device conceptualization stage to optimize a design so that it conforms to the anatomies of the patient population, leading to better clinical outcomes.

Design V&V activities include several types of testing scenarios that ensure the safety of the design with regard to strength, wear, stability, locking mechanisms, MRI compatibility and many others. CM&S can also provide guidance for the optimal use of a product by surgeons. It can be used to compare designs—either predicates or competitive designs—to demonstrate the advantages (or disadvantages) of one design vs. another. Finally, CM&S can be utilized for postmarket evaluations of unforeseen situations.

Examples of Simulation in the Medical Device Life Cycle

An extensive digital anatomic library that captures ethnic and gender variation across the global population was used to develop an anatomically shaped tibial baseplate, optimizing the kinematics, impingement and fixation aspects.1 (See Exhibit 1.) While ASTM- and ISO-standard test methods are available to guide the strength testing of tibial baseplates and hip stems, such prescribed standards are not available for other devices such as total ankle replacement (TAR) components.2

Exhibit 1: Statistical shape analysis guides anatomical knee design

CMS 1

 

To address this test methods gap, a physiologically-motivated loading rationale was developed using literature data to provide load and motion profiles throughout the walking gait cycle. Based upon the predicted component stresses, a physical fatigue test setup was developed using the worst-case gait position. (See Exhibit 2.)

Exhibit 2: TAR gait simulation to inform laboratory testing

Figure2 notsohighres

In cases where literature data is not available, testing inputs can be generated from a musculoskeletal model using patient-specific data to generate forces and moments passing through various foot joints throughout the gait cycle. (See Exhibit 3.)3

Exhibit 3: Patient-specific foot model to measure loading through gait in the Cuneiform Osteotomy3

Figure3 highres

The same musculoskeletal modeling approach can be extended to provide surgical guidance by studying ways that variability in surgical technique and patient anatomy affect the loading input.In cases where a standard test method is available, the prescribed testing may not be physiologically relevant enough to generate sufficient confidence in the design.

One such example is contact area and pressure testing for TAR to mitigate the risk of polyethylene wear. While the standard test method evaluates designs at specific flexion angles in isolation, e.g. anterior-posterior translations and internal-external rotations, the current method does not account for either the combined effects of these motions or the expected physiological loading magnitudes during the gait cycle. The TAR modeling approach described in the previous paragraph was used to study the effects of bicondylar TAR design on contact area and pressure throughout the gait cycle. (See Exhibit 4.)4 Moreover, the same approach was used to study the effects of edge loading due to tibiotalar mal-alignment during surgery, as well as comparisons between designs.

Exhibit: 4. Contact pressure throughout gait to assist comparison between TAR designs5

Figure4 highres

Adequate bony support is considered essential to achieve primary fixation in TAR. With the advancement of computational technologies, CT scan databases can be used to generate evidence for guiding surgical procedures, which could improve clinical outcomes. As one example, a computational study used CT data from 116 subjects to measure bony support (average normalized Hounsfield unit and resection area) for the tibia and talus bones, using both the flat and round (anatomic) virtual surgery cuts. (See Exhibit 5.) Several bone cuts were analyzed at incremental resection depth levels. The results showed that for both the tibia and talus bones, at all the resection depths, bony support may be decreased using flat compared to round cuts.

Exhibit 5: Effects of bone cuts on TAR bony support6

Figure5 highres

The Regulatory Pathway for Computational Modeling

As shown in the previous section, CM&S studies are an integral part of the total product life cycle of orthopaedic implants that can provide significant benefits to the device manufacturer. However, uncertainty regarding the credibility requirements of computational model results in regulatory submissions has limited the use of CM&S evidence during the regulatory review process. FDA began to publicly address this regulatory uncertainty by establishing CM&S as a regulatory science priority in 2011.7–9 Also in 2011, Dr. Bill Maisel, FDA’s Deputy Director for Science, was quoted as saying, “[At FDA], we’ve been looking at computational modeling as a fourth-pillar of premarket evaluation…If we have high quality models, we can help better devices to be developed more quickly.”10 The first statement signaled a sea change in FDA’s opinion on the value of CM&S tools for medical device development and review, putting computational modeling on equal footing with other “models” of device evidence. (See Exhibit 6.)

Exhibit 6: Computational modeling has been identified as an additional source of evidence when establishing device safety and efficacy.

CMS 6

The second part of his remark highlighted the need for establishing the credibility of a computational model before relying on CM&S results for device evaluation and approval.

Since 2011, FDA has undertaken numerous initiatives to address the regulatory uncertainty associated with CM&S. These include the formation of the Medical Device Innovation Consortium, the development of guidance and standards focused on CM&S for medical devices,11 providing a framework for incorporating virtual patients into clinical trials,12 and organizing conferences and workshops focused on CM&S for medical devices. The following section reviews guidance and standards that enable the use of computational modeling results in device submissions.

Reporting Requirements

The FDA recently finalized guidance entitled “Reporting of Computational Modeling Studies in Medical Device Submissions.”11,13 This document addresses one critical regulatory hurdle by recommending formatting and content guidelines for CM&S results that are included as scientific evidence in a device application. Another goal of the guidance is to improve the “consistency and predictability” in the review of computational modeling studies.The guidance describes a CM&S Report, where the suggested format is general enough such that it could be applied to most physics disciplines. (See Exhibit 7.) This includes reporting key model elements such as geometry, material properties, boundary conditions and solution details.

Exhibit 7: CM&S Report outline13

CMS 7

The guidance also includes a series of appendices that provide more detailed reporting recommendations for computational fluid dynamics, solid mechanics, electromagnetics, ultrasound and heat transfer.  For example, the appendix reviewing “Computational Fluid Dynamics and Mass Transport” recommends that the user provide information about the governing equations utilized in the model, e.g. the Navier-Stokes equations or Darcy’s Law, the turbulence modeling approach, and other auxiliary models, such as user-defined code.

V&V are highlighted in FDA’s reporting guidance as important components of the report; however, V&V processes are not described in detail. Instead, the guidance refers to existing standards, e.g. ASME V&V 10,14 ASME V&V 20,15 and IEEE 1597.1-2008.16

Establishing Model Credibility through VVUQ

Verification, validation and uncertainty quantification (VVUQ) establish the credibility of a computational model for a specific context of use.  Briefly, the goal of verification is to “assesses the numerical accuracy of a computational model” and the goal of validation is to “assess the degree to which the computational model is an accurate representation of the physics being modeled.”14  Uncertainty quantification seeks to quantify the uncertainties in the simulation results that are due to inherent variability in, or lack of knowledge about, the model input parameters.

Existing standards establish V&V practices that are readily applied in situations that permit significant data collection during physical testing in the intended use environment, e.g. wind tunnel testing of a wing design or crash testing a car.  But knowledge and application gaps must be addressed before rigorous V&V can be applied by the medical device community.  These include a lack of knowledge about key model input parameters, such as the naturally occurring shape variability of anatomical structures, the in vivo responses of living tissues and physiologic boundary conditions and loading modes. Our inability to interrogate the numerous multi-scale (physiological) interactions between an implanted device and the human body also limits model accuracy. In contrast, the increasing desire of the medical device community to utilize computational modeling in the development process places pressure on existing regulatory frameworks.

The ASME V&V 40 Sub-Committee for Computational Modeling of Medical Devices (V&V 40) formed in 2011 to address this regulatory need.17 V&V 40 sub-committee members include medical device companies, industry service providers, commercial software developers and FDA. The first work product of this sub-committee is a risk-informed credibility assessment framework (See Exhibit 8.), which establishes a link between the risk associated with using a computational model for decision-making and the model credibility requirements. The connection between model risk and model credibility is a foundational tenet of the V&V 40 standard that enables users to determine and justify the appropriate level of credibility required to use a computational model for decision-making.

Exhibit 8: ASME V&V 40 risk-informed credibility framework flowchart13

Figure8 highres

 

The process begins by posing a question of interest (QOI), i.e., the reason for the computational investigation.  The user then defines a context of use (COU), which is a statement about how the computational model will be used relative to other evidence. The next step is to assess the model risk, which is the risk associated with the decision being made based on the computational model results. The risk assessment drives the selection of the model credibility requirements, where a low risk scenario has lower credibility requirements versus a high-risk scenario. The V&V plan is established and executed once the credibility requirements have been determined. The next step is to assess the credibility of the computational model. The user then documents the model if it is deemed credible, or may revisit the risk-informed credibility framework if the model does not meet the established credibility requirements.

Locking Mechanism Strength in Total Knee Arthroplasty (TKA)

A finite element analysis (FEA) model to measure the strength of a TKA device locking mechanism is utilized to demonstrate the application of portions of the V&V 40 credibility framework. The QOI is whether the locking mechanism between the tibial insert and tibial baseplate of a posterior-stabilized (PS) TKA has sufficient strength to withstand posteriorly-directed loads.

Exhibit 9 presents cross-sectional views of the TKA components, illustrating the anterior liftoff of the tibial insert from the tibial baseplate when subjected to a posteriorly-directed load. There can be multiple COUs for the FEA model based on what additional information is available. For example, one COU (COU1) could focus on evaluating the anterior liftoff using the FEA model exclusively. A second COU (COU2) could evaluate anterior liftoff of a new and a predicate design using FEA, followed by benchtop evaluation of both designs. For COU1, the FE model results are the sole factor for decision-making, while the model results are a minor factor for COU2. And while the decision consequence remains unchanged for both COUs, the model influence, and hence the model risk, is higher for COU1 versus COU2. Therefore, the model credibility requirements are increased for COU1.

Exhibit 9: Zimmer Biomet TKA – Anterior Liftoff Setup13

CMS 9

The V&V 40 standard also outlines credibility factors, which divide model credibility activities into a set of constituent components. The goals for each factor are planned differently based on the model risk associated with the COU. A few of these credibility factors (highlighted in yellow in Exhibit 10) are now briefly described.

Exhibit 10: Credibility factors of CM&S V&V activities13

Figure10 highres

Model Form: One of the key assumptions of a computational model is the functional form of the material model, e.g. the model used for the polyethylene material of the tibial insert. Out of several material models available in the literature, there may be more flexibility and less need for justification in low risk situations (i.e. COU2). But for COU1, it may be necessary to develop a validated material model and quantify its impact on model predictions.

Model Input: One of the input parameters to the computational model is the geometry of the locking mechanism region, which is affected by the tolerances of the manufactured components. Parts may be modeled at nominal dimensions in lower model risk scenarios (COU2). Sensitivities of least and most material conditions (LMC, MMC) may need to be incorporated for higher risk scenarios (COU1).

Equivalency of Input Parameters: The loading could be applied using tibiofemoral load patches on the tibial insert for COU2. Including a femoral component to apply load through its sliding interaction with tibial insert may be necessary for COU1.

Conclusions

Computational modeling is used extensively throughout the product life cycle of orthopaedic implants and other devices.

Its use is not limited to simulating bench testing, but also to drive the creation of advanced and relevant test methods. Due to the advancement in computational technologies, CM&S is expanding to several new disciplines, such as MRI labeling, morphological analyses, patient-specific modeling and additive manufacturing.  But regulatory uncertainty regarding the validation requirements of computational model evidence has hampered the utilization of CM&S in regulatory submission process. An FDA guidance document is already available that establishes reporting best practices of modeling studies in device ubmissions.  A standard to guide the development of risk informed credibility requirements of computer models is also nearing completion. And since medical devices are developed for the global population, efforts are also ongoing to expand these V&V efforts by involving regulatory bodies outside the United States, enabling this regulatory pathway to be utilized around the world. Looking to the future, in silico clinical trials based on large datasets of digitized patients could usher in a new era of highly innovative products, reducing the industry’s reliance on in vitro and in vivo testing while maintaining (or even improving) device safety.


Mehul Dharia worked in the area of computational modeling in a wide variety of industry sectors, including automotive, defense, power, nuclear and consumer products, before joining the computational biomechanics research group at Zimmer Biomet in 2002. During his time at Zimmer Biomet, he has been involved with modeling various orthopaedic implants including hips, knees, shoulders, foot & ankle and trauma. Specifically, he has focused on the modeling of standards-based test methods, as well as development of physiologically motivated pre-clinical test methods to ensure the safety and efficacy of emerging implant technologies. He can be reached by email.

Marc Horner, Ph.D., is the Lead Healthcare Specialist at ANSYS, which he joined after earning his Ph.D. in Chemical Engineering in 2001. He began by providing support and professional services for biomedical clients, primarily in the areas of cardiovascular devices, drug delivery, packaging, microfluidics and orthopaedics. During this time, he developed numerous modeling approaches that can be used to establish the safety and efficacy of medical devices. He can be reached by email.

References

[1] Y. Dai and J. E. Bischoff, “Comprehensive assessment of tibial plateau morphology in total knee arthroplasty: Influence of shape and size on anthropometric variability,” J. Orthop. Res. Off. Publ. Orthop. Res. Soc., vol. 31, no. 10, pp. 1643–1652, Oct. 2013.

2] M. A. Dharia, “Biomechanical Evaluation of Total Ankle Arthroplasty,” in Proceedings of World Congress of Biomechanics 2014, Boston, MA, 2014, vol. 14–IS–3788–WCB.

[3] M. A. Dharia, J. E. Bischoff, J. Woodburn, S. Telfer, and A. A. Al-Munajjed, “Influence of Patient and Surgical Variability on Loading Across a Cuneiform Osteotomy,” in Proceedings of Orthopaedic Research Society Conference 2015, Las Vegas, NV, 2015.

[4] M. A. Dharia and J. E. Bischoff, “Walking Gait Simulation of Total Ankle Replacement Prostheses,” in Proceedings of Orthopaedic Research Society Annual Meeting 2014, New Orleans, LA, 2014, vol. 39, p. 1937.

[5] M. A. Dharia, “Edge Loading in Total Ankle Replacement – Does Bearing Type Matter?,” in Proceedings of American Orthopaedic Foot & Ankle Society Annual Meeting 2014, Chicago, IL, 2014, vol. e19.

[6] J. E. Bischoff, L. Schon, and C. Saltzman, “Impact of Anatomical Cuts on Bony Support in Total Ankle Arthroplasty,” in Proceedings of Orthopaedic Research Society Conference 2016, Orlando, FL, 2016.

[7] “Regulatory Science in FDA’s Center for Devices and Radiologic Health: A Vital Framework for Protecting and Promoting Public Health,” 2011.

[8] “US FDA CDRH Regulatory Science Priorities (FY 2016),” Oct. 2015.

[9] “US FDA CDRH Regulatory Science Priorities (FY 2017),” Nov. 2016.

[10] “FDA: Computational Modeling Could Lead to Big Gains in Device Safety,” MDDI Online, Nov. 2011.

[11] US Food & Drug Administration, “Reporting of Computational Modeling Studies in Medical Device Submissions: Guidance for Industry and Food and Drug Administration Staff,” Sep. 2016.

[12] T. Haddad, A. Himes, L. Thompson, T. Irony, R. Nair, and MDIC Computer Modeling and Simulation Working Group Participants, “Incorporation of stochastic engineering models as prior information in Bayesian medical device trials,” J. Biopharm. Stat., pp. 1–15, Mar. 2017.

[13] T. Morrison, M. A. Dharia, and M. Reiterer, “FDA-sponsored Training Seminar,” presented at the 2017 BMES/FDA Frontiers in Medical Devices Conference, Hyattsville, MD, 2017.

[14] “Guide for Verification and Validation in Computational Solid Mechanics,” ASME V&V 10, 2006.

[15] “Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer,” ASME V&V 20, 2009.

[16] “IEEE Standard for Validation of Computational Electromagnetics Computer Modeling and Simulations,” lEEE Std 1597.1, 2008.

[17] “ASME V&V 40 Verification and Validation in Computational Modeling of Medical Devices.” 

 

Join us!

The best of BONEZONE content delivered to your inbox, twice each month.

RELATED ARTICLES



CONTACT BONEZONE

 

CONTACT BONEZONE