Toxicologic Pathology Forum*: Opinion on Assessing and Communicating Adversity for Implantable Medical Devices

Abstract

Medical devices are a product class encompassing many materials and intended uses. While adversity determination is a key part of nonclinical safety assessments, relatively little has been published about the unique challenges encountered when determining adversity for implantable medical devices. The current paper uses the Society of Toxicologic Pathology (STP)’s “Scientific and Regulatory Policy Committee Recommended (‘Best’) Practices for Determining, Communicating, and Using Adverse Effect Data from Nonclinical Studies,” which were crafted for conventional bio/pharmaceutical products (small and large molecules, cell and gene therapies, etc), as a framework for making adversity decisions for medical devices. Some best principles are directly translatable to medical devices: (1) adversity indicates harm to the animal; (2) effects should be assessed on their merits without speculation regarding future or unmeasured implications; (3) adversity decisions apply only to the test species under the specific conditions of the nonclinical study; and (4) adversity decisions and supporting evidence should be clearly stated in reports. However, unique considerations also apply for evaluating implanted medical devices, including testing of multiple articles in the same animal and the unavoidable tissue trauma during device implantation. This opinion piece offers suggestions for applying previously published STP best practice recommendations for assigning adversity to implantable medical devices.

*This article is an opinion piece submitted to the Toxicologic Pathology Forum (TPF). This perspective is the particular view of the authors and does not represent an official position of the Society of Toxicologic Pathology (STP), British Society of Toxicological Pathology (BSTP), or European Society of Toxicologic Pathology (ESTP), nor should it be considered to reflect the opinions, policies, or positions of the authors’ employer/organization or regulatory agencies. The TPF is designed to stimulate discussion of topics relevant to regulatory issues in toxicologic pathology. Readers of Toxicologic Pathology are encouraged to send their thoughts on TPF opinion articles or ideas for new discussion topics to [e-mail].

Introduction

Medical Device Background Information

As defined by Section 201(h)(1) of the US Food, Drug, and Cosmetic Act, ⁵ medical devices work to treat/diagnose/prevent disease via a primary mechanism that does not require chemical reactions or metabolism to work. In the United States, medical device categories are defined in the Code of Federal Registration (CFR), where the definition for each device includes a CFR code, a name and risk designation (as class I [low risk], II, or III [high risk]), and a description. Similar classifications and systems are defined and regulated by other countries. ²⁷ A completely novel product that does not fit within a current CFR code is by default a class III device, although it can and often does move into a lower category as more information about its risk becomes available. This regulatory structure means that the same base material (eg, titanium), if previously used in a class I or II device with defined tissue effects, may be used in many different devices across multiple risk classes. “Clearance” is the Food and Drug Administration (FDA) process whereby safety and effectiveness of the test device are compared to establish equivalence of the novel product to a predicate device (ie, an already approved and marketed device with comparable physicochemical properties).

The US FDA recognizes five, often overlapping steps for medical device development. ⁶ These steps include: (1) device discovery and concept; (2) nonclinical research—prototype; (3) pathway to approval; (4) FDA review; and (5) post-market safety monitoring. While class III devices must demonstrate safety and efficacy (and are approved) on their own merits, class I and II devices (which comprise most of the devices reviewed by the FDA) assess safety/efficacy of novel devices in comparison to predicate devices in the same product category. Clearance and approval can be supported by nonclinical and clinical studies. However, because clinical testing in healthy volunteer subjects is not relevant with medical devices, all nonclinical studies are generally completed before any clinical studies are initiated which is why this paper uses the term “nonclinical studies” to describe animal studies that occur in preclinical stages of development.

Due to the need for comparative nonclinical studies to clear most class II devices, control groups are a key component of nonclinical medical device evaluation. Reference (positive control) materials that incite a known spectrum of tissue reactions, such as United States Pharmacopeia (USP) reference high-density polyethylene, are used to comparatively assess for local tissue trauma at the implantation site. Negative controls (no expected tissue reaction) and sham controls (typically the surgical approach without device implantation) can aid in assessing tissue reactions related solely to the procedures, including the extent of tissue trauma or contamination of the implantation site. Predicate controls are used to detect if the tissue reaction to the test device is similar to the expected responses deemed acceptable for an already marketed comparator device.

In recent years, an increasing number of publications in the toxicologic pathology literature have been dedicated to medical device evaluations. In this regard, landmark contributions include a special issue in the journal Toxicologic Pathology (Volume 47, Issue 3, April 2019) on various topics related to medical devices and a recent “Points to Consider” paper that outlines key pathology elements to tackle in evaluating the safety of implanted medical devices. ²⁰ This opinion piece discusses a topic that was not addressed in detail in these prior publications, namely specific approaches to assessing and communicating adversity for medical devices. Since medical devices are a very broad category of test articles, we have limited the scope of this current discussion to local and systemic evaluation of implantable devices that are not combination products (ie, products incorporating both a device and a primary bioactive component [“drug”]), and specifically how anatomic pathology endpoints drive adversity decisions for such test articles. Implantable devices are surgically placed in tissue and include nondegradable products that penetrate tissue and remain within the body permanently or for some time before being removed (eg, vascular balloon catheters) as well as degradable devices which may undergo metabolism/degradation but are not required to undergo degradation to achieve their primary mechanism of action. Importantly, medical device regulations also govern approval of in vitro diagnostic tests, but these devices are not considered in this paper since they generally do not require nonclinical testing or adversity decisions.

Challenges When Applying Nonclinical Adversity Practices for Conventional Bio/Pharmaceuticals to Implantable Medical Devices

Adversity in animal studies has been defined simply as “harm to the test animal” ¹⁸ or more elaborately as diminishment of the “functional capacity to maintain homeostasis and/or an impairment of the capacity to respond to an additional challenge.” ²¹ A weight of evidence approach is typically recommended when considering changes within the specific nonclinical study and test system (ie, animal model), without regard to the intended use of the test article or potential translatability of findings for predicting responses in other species (including humans). Recent publications have discussed general principles for determining adversity in nonclinical studies^{1,3,8,17,19,22,24} with more specific guidance for special situations including clinical pathology, ²⁵ thyroid alterations, ¹² immunomodulatory test articles, ²³ and direct central nervous system administration of oligonucleotides. ² Medical devices represent an additional unique product category when approaching adversity designations in the nonclinical setting.

The current Society of Toxicologic Pathology (STP) best practice recommendations for assessing and communicating adversity in nonclinical studies ¹⁸ were “written mainly in reference to bio/pharmaceutical perspectives,” but “many of the recommendations address fundamental views that also apply to safety assessments for environmental chemicals, food, and medical devices.” That said, the nature and therapeutic actions of non-combination implantable medical devices differ fundamentally from those of pharmacologically active test articles, thus requiring adjustment in applying the STP’s adversity recommendations. Our experience with medical device evaluation permits us to highlight several factors in both study design/structure that often complicate adversity assessment for implantable medical devices.

Biocompatibility rather than adversity is historically used to define medical device safety

Biocompatibility (biological compatibility) with its positive connotations, rather than adversity with its negative attribution, is the historical determinant used to establish safety of implanted medical devices. Biocompatibility is generally defined as “the ability of a device material to perform with an appropriate host response in a specific “situation.”” ⁷ Historically, medical device standards including biocompatibility were developed within non-biological (mainly material science and engineering) disciplines, with a focus on evaluating local changes (ie, at the device-tissue interface) with limited investigation of systemic effects. Although biocompatibility remains the main safety measure for medical devices specified in extant standards ¹⁵ and in published literature, ²⁰ the FDA has recently begun to use adversity as their approach for biological assessment of medical devices (FDA.gov).^4,7

Confounding tissue injury commonly occurs with medical device implantation (ie, localized tissue “harm” is required to place most devices)

Some degree of tissue disruption is unavoidable when most medical devices are implanted. The adversity of such procedural effects depends on several factors including the size of the device and the tissue into which it is introduced. For example, neural tissue disruption is an intrinsic consequence when a rigid material is placed in the brain (eg, indwelling cannula or neurostimulatory device) or introduced one or more times into a narrow cerebrospinal fluid–filled cavity (eg, catheter or injection needle). This intrinsic tissue damage is why control groups for nonclinical studies of novel medical devices often include sham/positive/reference/predicate controls to discriminate procedure-induced injury from any test article–related effects. For this reason, adversity decisions for medical devices are regularly complicated by the need to separately assess the extent of tissue injury caused by the test article and procedure. In contrast, for conventional bio/pharmaceutical products, test article–related “harm” can usually be segregated from procedural effects with comparative ease since injected agents are administered by narrow catheters or needles causing minimal tissue injury while other routes of administration (eg, inhaled, oral) cause essentially no damage. We recognize that some chemically/metabolically functional test articles administered locally or surgically into specific tissues compartments (eg, skin, brain, eye, or ear) may also induce substantive procedurally induced local tissue injury.

Reversibility/cessation of treatment may be challenging or impossible to demonstrate for many implanted medical devices

In general, the term “reversibility” applies to externally applied or communicating devices (eg, patches). However, “reversibility” may infrequently be applicable to internally implantable devices. This situation reflects the fact that implanted devices are either permanently placed or the surgical procedures may lastingly alter the tissue organization at the implant site.

For implantable devices, the closest analog to reversibility (as assessed for bio/pharmaceutical products) may be bioabsorption. For biodegradable devices (which are designed to dissolve over time), the extent of bioabsorption (“recovery”) typically needs to be determined. Biodegradable devices often incite a robust local immune response until most of the material has been removed from the site and the tissue response settles at a new steady state, though generally not returning completely to the pre-implantation baseline. One commonly requested evaluation for biodegradable devices is staging the extent of device degradation or degree of bioabsorption. ⁷ Factors including the physicochemical nature of the biomaterial, type and activity of the device, and the implant location influence whether complete bioabsorption occurs during the time frame of an individual study. Notably, it is common for small (ie, biologically inconsequential) amounts of device biomaterial and a residual population of immune cells (commonly lymphocytes as sentinels and long-lived macrophages) to remain within an implantation site that has otherwise regained functional and structural tissue integrity after the bulk of the device has been degraded.

Dose (and by extension a “No Observed Adverse Effect Level”) is typically not relevant for devices

In general, implantable medical devices have a single “dose” unit, with some scaling for differences in body size (eg, pediatric vs. adult bone plates). Implantation of different amounts of test article (ie, numbers of a test device) is not common for devices that lack a chemical/metabolically functional component, just as implantation of a partial device is not common. Moreover, multiple biomaterials may be present in an individual device, the bulk of which are biologically inert, so the total amounts and kinds of material are often much less important than the location and associated tissue injury induced by implantation. Additionally, neither overdoses nor supraphysiological effects are concerns for implantable medical devices lacking a chemical/metabolically functional component. Instead of dose, the length of time a device will be in contact with the body is more important to the assessment, for example, the intended length that an indwelling catheter can be used before being removed. This is typically categorized as: (1) limited or <24 hours; (2) prolonged or 24 hours to 30 days; and (3) permanent or more than 30 days. High-risk devices, particularly those that replace a critical biological function (eg, implanted defibrillator or ventricular assist device), make graded dosing or assessment of partial devices irrelevant and potentially lethal if using a partially functional device. Therefore, adversity decisions for implantable devices will need to be supported by attributes other than a conventional threshold value such as a No Observed Adverse Effect Level (NOAEL).

Study designs for implantable devices often include multiple test articles/interventions in the same subject, which can prevent assigning a systemic tissue change to a specific test article

Many nonclinical studies of medical devices employ study designs that include multiple implantation sites/treatment interventions in the same animal. For example, a wound healing study may include 6 to 8 skin sites per animal, with negative control, predicate, and test article treatments each applied to 2 to 3 sites per animal. This approach is particularly true for non-rodent test systems (eg, rabbits, dogs, minipigs, nonhuman primates) where larger body sizes can accommodate multiple spatially distant implantation sites and can reduce the number of animals required on study, consistent with ethical considerations and the 3R (“reduce, refine, replace”) principles. ¹¹ While it can be possible to determine if a local tissue reaction is more severe for one implanted device than another, the presence of multiple implant types in a single animal can make it impossible to assign a specific cause to systemic findings.

Nonclinical studies of medical devices often combine efficacy and safety assessments

To initiate clinical trials, nonclinical programs for medical devices require assessments of both safety and efficacy with the “final finished form” (FFF), that is, a medical device or device component that includes all manufacturing processes for the “to be marketed” device including packaging and sterilization, if applicable. ⁷ In many instances, the efficacy and toxicity assessments may be combined in a single pivotal subchronic or chronic study. In some cases, endpoints for efficacy and toxicity require separate samples, which may not be feasible for small devices. The need for separate samples can complicate the integrated interpretation of efficacy and toxicity responses in nonclinical studies.

Nonclinical studies of medical devices often utilize animal models of disease using species or strains with limited historical databases regarding spontaneous background findings

For medical devices, efficacy studies as standalone bioassays or in combination with toxicity testing may be performed in spontaneous or induced (eg, by genetic engineering or pharmacological or surgical intervention) models of disease that mimic the conditions of the clinical indication. These animal models of disease are not healthy subjects and will often exhibit comorbidities that can confound interpretation of adverse findings related to any implanted medical device. Moreover, such disease models may rely on less common laboratory animal species (eg, guinea pigs, mongrel dogs) that often lack historical databases of age-related and spontaneous lesions that can be used to assist in interpreting the adversity of device-related findings. Similarly, some species used in device studies are not readily available from typical laboratory animal suppliers and require special housing conditions (eg, calves, conventional swine, sheep), which may impart additional variability in biological responses due to unusual environmental conditions and husbandry practices. The paucity of background data may be a substantial impediment to interpreting efficacy and toxicity endpoints in unusual models.

Appropriate control materials for testing medical devices are often difficult to identify

Vehicle controls consisting of excipient components are frequently used as comparators with pharmaceuticals, and saline may be the appropriate excipient control for biologics. Conversely, medical devices are generally compared to a positive control material (ie, induces a previously characterized [known] tissue reaction), such as reference control material (eg, USP reference polymer), or to a predicate (marketed) device. Truly negative controls or saline controls are not generally appropriate, but a sham surgical control may be useful to interpret procedure-related changes. The comparative positive/reference/predicate control materials may have different physiochemical characteristics and induce different cell and tissue changes (eg, extent and type of inflammation) and may have a different temporal inflammatory and/or degradative sequence compared to the test materials. In many instances, the commonly used positive or reference controls act more like surgical procedure controls and are less appropriate as a direct comparator to the inflammatory or other tissue responses induced by the test medical device. Additionally, unless the test and control materials are implanted in the same location that will be used for the clinical indication, reactions to the test and/or control materials may not predict the expected clinical response. Ideally, the clinically expected location is recapitulated in the nonclinical studies, but the size of the FFF device, anatomical differences between animals and humans, and/or the need to reduce surgical mortality (eg, peripheral vein as a surrogate for cardiac vein to model invasive cardiac vein engraftment) may not allow use of the clinically expected location for all nonclinical studies. ²⁶

Medical devices can fail mechanically due to design, surgical and user errors, and biotribology forces

In vivo failure of a medical device, acutely or chronically, may lead to a variety of sequelae ranging from minimal (nonadverse) to adverse life-threatening events requiring replacement (revision) and additional surgeries or can result in a lethal event. Engineering risk management includes a proactive failure mode and effects analysis for design, process, application, and service of the product and to identify potential hazards and risks as well as ways to reduce them. ¹⁶ Despite this analysis, unexpected design or material flaws can lead to unexpected or premature physiochemical damage to the device (eg, “fracture” of artificial heart valve). Unexpected or inappropriate surgical procedures, inadvertent microbial contamination of the device or surgical site, or inappropriate use of the device can also contribute to partial or complete failure. Finally, devices intended to undergo motion in the body (eg, prosthetic joints) are subject in the long term to friction and wear as well as lubrication limitations that can contribute to biotribological failure of implants (where “biotribological” refers to the impact of the study of mechanical forces in biological systems). Except for post-surgical infections, these physical and functional failures of devices may be difficult to identify in tissues from animal safety studies. Unexpected mortality evaluations, careful examination of the implantation site, and evaluation of potential tissue sites for embolic deposition can identify some evidence of adversity, but clinical signs, imaging, and functional testing with an integrated weight-of-evidence approach are usually necessary to confirm partial or complete device failure.

Opinion and Suggestions for Assigning and Communicating Adversity for Nonclinical Device Studies

Given this background information, we are now positioned to discuss how to apply the ten existing STP “best practice” recommendations for determining and communicating adversity decisions in nonclinical studies ¹⁸ to medical devices. Briefly, STP’s published recommendations (1-10) are given below and our opinion/suggestions are provided underneath each recommendation. Recommendations 1 through 4 address generic points related to determining adversity. Recommendations 5 and 6 focus on communicating adversity. Recommendations 7 and 8 are focused primarily on establishing a NOAEL. Recommendations 9 and 10 focus on collaboration between experts in determining risk.

For clarity, the STP’s published adversity recommendations are given below, and our suggestions regarding their applicability (or lack thereof) to implantable medical devices are provided underneath each recommendation. In general, STP recommendations 1 to 6 and 9 and 10 have the most direct applicability to nonclinical studies of implantable medical devices. In contrast, recommendations 7 and 8 (which address NOAEL decisions) are not relevant to most implantable devices. For each of the ten STP recommendations, we state the extent to which we agree that the principles can be directly applied to adversity decisions for implantable devices, describe any additional caveats for using the recommendation for nonclinical studies of devices, and/or discuss why the recommendation has minimal/no applicability to most nonclinical device studies.

Determining Adversity—Recommendations 1 to 4

Recommendation 1: “Adversity Is a Term Indicating Harm to the Test Animal”

Defining “harm” is a continual challenge in adversity assessment, and nonclinical studies of medical devices often have additional challenges in determining harm compared to bio/pharmaceutical test articles. We see this recommendation as being fully applicable to medical device studies, subject to certain additional factors for consideration.

For chemically/metabolically functional test articles administered orally or by a single parenteral injection, an appropriately performed dosing procedure typically produces negligible or no tissue injury on its own (unless injected or surgically introduced into a specific tissue compartment such as skin, brain, ear, or eye). In contrast, local tissue injury when implanting medical devices is unavoidable because the surgical approach is inherently traumatic, and the severity of trauma can vary widely based on the device type (especially size) and the anatomy of the implantation site. The interpretation of medical device–related effects is often further complicated if multiple devices are placed in the same animal or if control conditions are limited or non-existent (eg, no sham or predicate control sites to show the baseline trauma caused by a procedure).

At this time (based on literature review and discussions within the author group and peers with substantial experience evaluating devices), no consensus answers appear to exist for key questions: (1) Is unavoidable procedure-induced injury to sensitive tissues (eg, death of terminally differentiated retinal cells when inserting a retinal implant) an adverse change?; and (2) Alternatively, since tissue injury cannot be avoided when implanting the device, is the change nonadverse as long as the injury is limited to the minimal damage necessary to place the device? In general, the STP approach ¹⁸ to these questions is to assign adversity to test article–related findings that exceed responses induced by control articles but not to attribute adversity to procedure-induced changes. However, our experience with implantable medical devices suggests that “procedural adversity” may be extensive at the implantation site and thus also deserves consideration independent of effects related to the implanted material.

Until there is a consensus on whether procedural manipulations that cause tissue injury are inherently adverse in their own right, we suggest that the pathology report for medical device studies clearly states which changes are interpreted to be adverse, and the rationale for determining adversity. This rationale could include considering any substantial injury as adverse (even if present in control/sham/comparator groups), separating procedure-induced changes (eg, “procedural” adversity) from test article–specific changes (“regular” adversity), or some other justification.

Recommendation 2: “The Decision About Whether or Not Test Article–Related Effects (or a Group of Related Effects) in a Nonclinical Study Are Considered Adverse or Nonadverse Should Be Unambiguously Stated and Justified in Subreports and/or the Study Report”

This recommendation is fully applicable to medical devices, with additional caveats to consider for implantable devices as described in the various sections above.

Some design features for nonclinical studies of implantable medical devices create unique challenges when trying to definitively assign adversity to a specific test article. Challenges include (but are not limited to) use of fewer animals per treatment; use of non-traditional animal species (eg, small ruminants that have lived on pasture rather than in facilities with carefully controlled environmental conditions and husbandry practices); and the use of multiple, often different implants in the same animal which can preclude the assigning of findings to a specific article/intervention. Additionally, the challenges discussed for recommendation 1 about inherent tissue damage during implantation also apply here.

If a nonclinical medical device study has higher ambiguity (due to one or more of the above discussed factors), we suggest that the report should state conclusions with as much certainty as possible. If the level of certainty is low, the confounding factors should be described (eg, low animal number, unusual test species or husbandry conditions, absence of applicable historical control data for presumed rare findings). For adversity decisions, the criteria (eg, responses to control conditions, if any) used to distinguish procedure-induced tissue injury from device-related injury should be stated, and factors considered in ascribing adversity to the implanted device versus the procedure should be defined in some detail.

Recommendation 3: “Adversity as Identified in a Nonclinical Study Report Should Be Applied Only to the Test Species and Under Conditions of the Study”

This recommendation is fully applicable to nonclinical studies of medical devices. The recommendation contains three main points.

Recommendation 3a: “When toxicity in a test animal is interpreted as being specific to that species and lacking relevance to humans, the test article effect may still be an adverse response for the species being tested.”

This point is well recognized for many chemicals including some bio/pharmaceuticals, where substantial injury in animals is not relevant for humans. Examples include male rat-specific renal tumors related to chemically induced α2u-globulin nephropathy ¹⁰ and rat-specific thyroid tumors associated with chemically induced hepatic enzyme induction. ⁹ In contrast, implantable devices with no chemical/metabolically functional component may have more direct predictive value for human risk assessment compared to biologically active molecules since devices do not work through chemical means or need to involve metabolism. Species-specific effects of potential relevance to predicting human responses are more likely for devices incorporating xenograft materials.

Recommendation 3b: “Neither the therapeutic indication nor the patient population should influence adversity decisions in the test species.”

We agree that the intended patient population does not impact adversity decisions for devices in nonclinical species, which is the case for other biomedical product classes. While we agree in the abstract that therapeutic indication should not be taken into consideration, the way in which devices were historically cleared can make it challenging to completely separate a device from the therapeutic indication. In general, recently cleared devices have details and specific instructions for use and will only be used for the listed indication or tissue and patient population (ie, are seldom if ever used off-label). However, many older devices were cleared with a very general indication (eg, “may be used wherever temporary wound support is required”). A device cleared with a general indication could then be used as a comparator in an anatomic location where it was never previously tested. Therefore, when clearance of a medical device is extended to another anatomical site and indication, the therapeutic indication may become relevant. For example, mesh materials initially cleared for inguinal hernia repair were not deemed adverse in nonclinical studies and clinical application, but extension (without animal studies) to use as transvaginal mesh for urogenital pelvic floor disorders induced clinical adversity (see supplement in O’Brien et al ²⁰ ). Similar adverse responses were shown retrospectively in animal studies and were at least partially due to biomechanical differences in the inguinal and pelvic region tissues.

Recommendation 3c: “Test article–related exacerbations of background lesions in animals can be considered adverse.”

A device may incite enhanced tissue reactions compared to control materials. For example, localized thrombosis around an indwelling venous catheter is common, ²⁸ but a much higher rate/severity for thrombosis for test catheters versus control catheters could represent a device-related enhancement/exacerbation of that common finding. Such enhancements are often quantifiable, but if they are fairly modest, they may be interpreted as biologically inconsequential.

Recommendation 4: “Toxic Effects on Cells, Tissues, Organs, or Systems Within the Test Animal Should Be Assessed on Their Own Merits”

This recommendation also is comprised of three main points, of which two recommendations 4b and 4c are fully applicable to nonclinical studies of medical devices (after considering the caveats listed below).

Recommendation 4a: “Effects believed to be suprapharmacological should be considered either adverse or nonadverse.”

Implantable medical devices with no bioactive component exert no pharmacological effect, so this point is generally irrelevant for non-combination products.

Recommendation 4b: “All relevant effects should be considered in the interpretation of adversity, regardless of whether they are perceived to be primary, secondary, or tertiary.”

Discrimination of primary (ie, direct test article–related) effects from secondary and tertiary effects (ie, indirect [downstream] changes not caused by the test article) is often feasible for conventional bio/pharmaceutical test articles but is typically not relevant to implantable medical devices. That said, the principle can be readily adapted to identifying adverse changes due to the surgical procedure (eg, injury to terminally differentiated retinal cells during placement of a retinal implant). The point provides no definitive recommendation regarding whether adversity for medical devices should be assigned to test article–related findings, procedure-associated injury, or the combination (alone or together), so each report author may make—but will need to justify—such determinations based on their own judgment.

Recommendation 4c: “Responses interpreted as being ‘adaptive’ should be considered either adverse or nonadverse.”

An adaptive response may be defined simply as a physiological adjustment that allows an organism to successfully accommodate continued exposure to a noxious stimulus (eg, prolonged xenobiotic exposure). Specific “adaptive” responses may differ for chemically/metabolically functional articles (eg, hepatic enzyme induction by bio/pharmaceuticals and other chemicals) versus implantable medical devices (eg, periosteal and endosteal woven bone formation to stabilize an implant). Nonetheless, the overall concept of adaptive responses is applicable to nonclinical studies for implantable medical devices.

Communicating Adversity—Recommendations 5 to 8

Recommendation 5: “Communication of What Is Considered Adverse and Assignment of the NOAEL in the Overall Study Report Should Be Consistent With, and Supported by, the Information Provided in the Study Subreports”

This recommendation contains four main points. We see the adversity portions of this recommendation (recommendations 5a, 5b, and 5d) as fully applicable to nonclinical studies of medical devices when considered with the additional factors for consideration listed below.

Recommendation 5a: “Test article–related changes should be documented in subreports, regardless of whether they are considered to be adverse or nonadverse.”

This statement is straightforward and needs no further explication. However, for medical device studies the discussion of tissue responses should include whether they are specific to the test device (including such key features as the device shape, size, and composition) or are related to the procedures needed to implant the device.

Recommendation 5b: “Test article–related adverse findings considered to be part of a constellation of related effects should be discussed together.”

This point is highly relevant for medical devices when assessing imputed systemic reactions to a locally implanted device.

Recommendation 5c: “The NOAEL is to be assigned for the study as a whole rather than to subreports.”

This point is not applicable to most nonclinical device studies because the NOAEL concept is seldom relevant except for combination products pairing a device with a drug. The lack of relevance is intrinsic to the nature of devices as such units are inserted but cannot be “administered” at multiple “doses” to fulfill the need to test the FFF.

Recommendation 5d: “Vague statements such as ‘not biologically relevant’ or ‘not toxicologically important’ should only be used in the study report when defined and supported by a sound rationale.”

Additional statements that would need to be defined in reports for medical device studies may include “not procedurally relevant” and “not anatomically relevant” (eg, humans lack the affected anatomic structure).

Recommendation 6: “Communication of Adverse Findings and the NOAEL Should Include Direct Interaction Between Staff Within Different Contributing Scientific Disciplines”

The adversity portion of this recommendation is fully applicable to nonclinical studies for medical devices. As noted above for recommendation 5c, the NOAEL portion is not applicable to most nonclinical device studies.

Recommendation 7: “The NOAEL for a Test Article Should Be Communicated in an Overview Document Based Upon Data From Multiple Studies”

Again, this NOAEL-centered recommendation is not applicable to most nonclinical device studies, for the reasons detailed above for recommendation 5c.

Recommendation 8: “In Order to Place Them in Appropriate Context, the Use of NOAELs in Data Tables Should Be Referenced to Explanatory Text”

This recommendation is also irrelevant to most nonclinical device studies, as explained above for recommendation 5c.

Using Adversity to Assess Human Risk—Recommendations 9 and 10

Recommendation 9: “Nonclinical Scientists, Including Toxicologists, Pathologists, and Other Contributing Subject Matter Experts Who Interpret Data From Nonclinical Studies, Should Be Active Participants in Assessing and Communicating Human Risk”

This recommendation is fully applicable to nonclinical studies of medical devices and requires no further embellishment. In our experience, once the contribution of pathologists as experts in assessing tissue responses is recognized, medical device pathologists are often valued participants in the decision-making process for assigning adversity since anatomic pathology endpoints are frequently central to such deliberations.

Recommendation 10: “All Available Data From All Nonclinical Studies Must Be Evaluated Together to Define Any Potential Toxicities and to Predict Human Risk”

This recommendation is fully applicable to nonclinical studies of medical devices. Some nonclinical studies for pharmacologically inert devices may have more direct predictive value in assessing human risk compared to chemically/metabolically functional test articles including combination products pairing a device with a drug. Based on this recommendation, biocompatibility of a medical device should not be determined until data from all in vitro and in vivo studies are available and can be integrated (ie, biocompatibility cannot be determined, nor should the test material be stated as biocompatible in any single study).

Additional device-specific considerations for adversity relevant to nonclinical device studies

Our collective experience has shown that the existing STP “best practice” recommendations for determining, communicating, and using adversity ¹⁸ constitute solid counsel for initiating adversity decisions, but the recommendations are not all-inclusive for the unique aspects of medical device studies. Accordingly, we offer these additional practical suggestions for determining adversity with respect to medical devices.

1. Adversity, rather than mere biocompatibility, should be the primary measure used in establishing safety when evaluating local tissue reactions (ISO 10993-6:2016 ¹⁵ ) and systemic responses (ISO 10993-11:2017 ¹³ ). Biocompatibility should be stated only after risk assessment based on compilation of all in vitro testing and determination of in vivo adverse findings has been completed.

2. Since implanted devices comprised of relatively inert materials (one or several) do not possess a biologically functional component (ie, working through chemical means or metabolic activity), the nonclinical development pathway typically does not need to assess enzyme or protein compatibilities across species. Pharmacokinetic data for such devices may only apply to in vitro evaluation of extractables and leachables (where an extractable is a substance released in the laboratory while a leachable is a substance released during use ¹⁴ ).

3. Potential toxicity of extractables and/or leachables may be a factor used to establish the device adversity, although such information is typically communicated in a document other than the pathology report.

4. Partial or complete physical failure of the device is often difficult to identify in tissue sections, but local and systemic changes need to be evaluated and used along with functional and other testing to identify and de-risk such adverse events.

5. Anatomic placement (eg, mesh implanted in an inguinal versus vaginal location), device size, duration of implantation, and species-specific anatomic features (presence/absence of certain organs, skin thickness, etc) may be substantive determinants of adversity in nonclinical device studies.

6. Determination of an adverse (or biocompatible) tissue response to a medical device cannot be based on a simple calculated difference between group mean semi-quantitative scores observed for test material and positive/reference/predicate control materials. Instead, adversity decisions for devices require integration of both qualitative and semi-quantitative parameters as well as knowledge of age-related and spontaneous findings expected for the test species and sex in each study.

7. Well-annotated “representative” images with detailed descriptions (ie, text and figure legends) for common and biologically consequential but less frequent tissue responses, all prepared by an experienced medical device pathologist, are much more relevant for communicating adversity and supporting risk assessment compared to mere comprehensive photodocumentation of tissue responses for all devices without accompanying expertly prepared descriptions. This point is particularly true if the person performing the risk assessment has little or no formal pathology training and limited access to expert pathology support to aid in interpreting the numerous images when performing the risk assessment.

Conclusions

This article is designed to offer practical guidance regarding how recognized “best practice” principles for determining adversity in nonclinical studies for other biomedical product classes may be effectively applied to implantable medical devices. Beneficiaries of our experienced viewpoint will include our medical device pathologist colleagues (regardless of their experience level); allied scientists (eg, chemists, engineers, toxicologists); and regulatory reviewers. We hope that this article will generate active discussion to help establish consensus perspectives on appropriate practices for assessing and communicating risk through adversity rather than biocompatibility determinations, of novel medical devices. In particular, we trust that this piece launches future papers to solidify the approach to adversity determinations for implanted non-combination devices as well as providing suggestions to shape adversity practices for use with other types of medical devices (eg, combination devices, devices in contact with intact skin/mucous membranes/cornea). Ultimately, the essential elements in assigning and communicating adversity in nonclinical studies of medical devices need to be clear and concise, which is no different from current best practice principles as applied to other types of biomedical products.