Biocompatibility testing approaches for medical devices

Transcription

1 Practical Application of the TTC Approach for Compounds Released from Device Materials Ronald Brown, US FDA, Center for Devices and Radiological Health Presentation In response to the questions related to inhalation, the potency values that the TTC values are based on are derived from oral toxicity studies. A number of investigators have looked at similar distributions of potency values for inhaled compounds. We re thinking along those lines for compounds released into gas pathway in a respiratory device. Relevant to Alan s question, what we re doing in the risk assessment is comparing dose to the patient to exposure levels. When we re talking about levels of a compound in a polymer (5 percent,.5 percent) we have to remember that not all of that may come out, depending on the polymer. Richard s example was a good rough cut of that. If you back calculate and there is no way that the patient could be exposed to levels of the compound above the exposure limit, even if it came out all in one day, then you re done. But often for devices that is not what happens; you ve got kinetics of release over time. Just something to keep in mind; you can t always use the percent in the polymer as the way to estimate patient dose. Traditionally, the first step would be to look for experimental and derived tox values for the compounds before shifting to the TCC approach. Now, we ll explore what the TTC really means and how we plan to use it. The TTC approach has been around a long time. At the FDA, it has been in use at least 20 years at the FDA, in relation to food additives. The ICH is currently thinking about how to use this approach to qualify genotoxic impurities in drug substances. It is always a challenge to borrow technology from one area and apply it to another, and that s especially true for medical devices. There are some key things to consider in a practical application of the TTC approach to devices. Biocompatibility testing approaches for medical devices There is truly a paradigm shift going on now. In the past, and to a large extent still, biocompatibility testing is typically done by taking an extraction of a material, and testing that in in vitro and animal models, or doing an implantation of that test. Now, we are trying to take advantage of advances in analytical chemistry, and evaluate an extract chemically instead of testing that extract biologically, identify all the chemical constituents present above a certain level, and then perform a risk assessment on each of those constituents. That sounds great on paper. But in reality we really do not have the toxicology data we need to do a risk assessment on each of these individual compounds. It is especially difficult when we cannot even find a CAS number for the compound; we can t uniquely identify it. The flow chart in the Part 1 ISO standard asks, Does sufficient toxicity data exist for all the compounds in the materials? More often than not, the answer is no, and we have to find a plan B and the standard does not supply clear direction at that point. At that point, when we don t have toxicity data for all the compounds that are being released from the device, the TTC approach has some merit as one alternative to consider. One practical way to use computational toxicology models to help predict the toxicity is to determine if the compound is likely to be carcinogenic or not. If we have data (mutagenicity data, predictions from QSAR models) suggests that it is carcinogenic, that can help us decide which TTC value to use. That is probably the best way to use the QSAR models to help us decide which TTC value to use cancer or not cancer-based. Use of TTC values in absence of compound-specific toxicity data

2 How are people thinking about TTC? Some people consider TTC as an approach to decide when compounds need additional testing or not. That sets the bar pretty low; it is a little bit different than saying that the TTC value can serve as a surrogate exposure limit. At the recent TC194 meeting in Pavia, Italy, participants considered the question of how the field can actually use the TTC approach. The medical device community is generally comfortable with the idea of using the TTC as a default exposure limit not just a way to decide whether you need testing or not, because medical device manufacturers are not going to test the individual compound. There are still some concerns about moving forward in that direction, but Brown said, I think it s a science-based approach, that has been around long enough and used in other fields, that we can feel comfortable using it as default exposure limits. The other way we can use it is as a cutoff for the limit of detection. If you ve got compounds released below the TTC value, there is little point in identifying them any further if they are not expected to be of toxicological concern. That s another way to use the TTC value as a practical way to do your chemistry evaluation. There are a number of challenges involved with using this approach for compounds that are released from medical devices. Let s explore and see if we can come up with proposed solutions. Challenge: Duration of exposure One challenge rests in how to adjust the TTC approach for less than lifetime exposure to compounds. For many medical devices, patients are exposed for a relatively short time. In those cases, how should we modify existing TTC values? The table below offers a simplified take on the ICH Draft Guidance for duration of exposure. TTC values for short and long-term exposure The ICH document only offers short-term exposure limits for carcinogenic compounds. The table above also provides TTC values for non-cancer effects, and these are stratified by Cramer classes. For non-cancer effects, compounds can be stratified in terms of potency and sorted into different categories based on their structure. In essence, compounds that have certain structural classes are likely to be more toxic than other types of structures. That concept is a pillar of toxicology, and Cramer class designation formalizes that. Still, short-term analogues for the non-cancer effects, the non-cancer TTC values are lacking. One solution is to use the tiered approach for short-term exposure that colleagues on the drug side are using, just to make it consistent. There is merit in trying to harmonize, and keeping the document for medical devices as similar to the guidance for drugs as similar as possible. Challenge: Route of exposure

3 Another challenge is whether or not these approaches are applicable for different routes of exposure; are the values adequately protective regardless of how patients are exposed? So many medical devices expose patients to compounds intravenously, or by other parenteral routes. The oral values are probably adequately protective for parenteral routes of exposure, but more research is needed to confirm that point, and ensure they are consistent with the ICH approach, outlined in the ICH M7 Draft Consensus Guidance (2013). According to that guidance, the ICH acceptable intakes for an individual impurity are applicable to all routes of administration. Given an unknown compound, with an unknown toxicity level, it is important to know that if a TTC limit is set, it will be adequately protective even for a proven potent compound, a really bad actor, for different routes of exposure. For a distribution cutoff, therefore, a conservative value is desirable, thus it is set at the 95 th percentile (thus, there is a probability of 95% that the NOEL value of the substance is higher than the TTC value). Because of the conservatism that was built into deriving the oral TTC values, colleagues on the pharmaceutical side believed that this would be protective for all routes of exposure. While intuitively that does not necessarily make sense, because some compounds are going to be a lot more potent by IV exposure than by oral routes, but down at low doses the difference is minimal. The ISO TC-194 working group agrees with the approach to use the ICH M7 proposed values across all routes of exposure. However, looking at the inhalation values separately might make sense. It would be ideal to have a distribution of inhalation potency values to draw from. Investigators at the Fraunhofer Institute 1 looked at their RepDose database, and did this type of analysis. Rather than automatically assume that the oral values are protective, they decided to look at a distribution of inhalation potency values, and see what the inhalation TTC values might be. This is a table from their paper, in which they compared three approaches: They derived values from the RepDose database, represented them in terms of ppm and mg/m 3, and then converted that to µg/per person/per day (the scale generally used in dealing with in the oral values, by assuming a certain breathing rate). They then compared those values to values that two other investigators came up with: A Carthew et al study focused on the inhalation route, and 1996 Munro et al. research with an oral route. 1 Escher, S.E. et al. Evaluation of Inhalation TTC values with the database RepDose. Regulatory Toxicology and Pharmacology. 58 (2) 2010: pp

4 In their analysis, for Cramer class 1 compounds they found a TTC value for inhalation of 180 µg/per person/per day. The Munro Cramer classification for non-cancer compounds was These are the values that are intended to be protective for non-cancer effects. So for the less potent compounds, the TTC value for Cramer class I, was 1800; for the more toxic compounds Cramer class III, it was 90 µg/day. For the potential carcinogens, it is 1.5 µg/day. When they do this analysis independently for Cramer class 1 compounds, they believe that the cutoff is tenfold to 15-fold lower than what we re seeing based on oral exposure for one of the more potent compounds. One of the challenges we face when we apply this kind of approach to device-related compounds is that often these exercises are done with very potent compounds (pesticides, all kinds of nasty things). In such cases, are the values derived for other data sets really applicable for the compounds that are likely to come out of devices? That is why we need more examples that use compounds known to come out of medical devices. Challenge: Applicability to mixtures, metal-containing compounds, and local effects (other biocompatibility endpoints) Is the TTC approach applicable for mixtures, and metal-containing compounds? A lot of the compounds in medical devices are either metallic or organo-metallic. Does this work for local effects? What about other effects we have to evaluate in that biocompatibility testing matrix? When you look at the Carthew values, interestingly they looked at local and systemic effects: Local effects in the respiratory tract, and then systemic effects as well and got somewhat different values. A lot of times investigators will not know if a compound will produce local or systemic effects, so to be conservative you can go with the lower values. As a reality check, consider a case study applying the different TTC values to compounds known to come out of respiratory devices. This case study uses data from two companies that make ventilators. Using data on actual volatile organic compound (VOC) levels that are measured at the patient port of their devices, a practical case study explored which type of TTC values for inhalation make the most sense. Case study: Application of TTC to VOCs released from respiratory devices In the example shown in the table above, two of the compounds, toluene and iso-butylacetate, are highlighted. Running it through the Tox Tree software program showed that Toluene was Cramer class I not very toxic. Exposure limits in airborne concentration are provided. This case study represents the default method when toxicology data on a compound is not available. If toxicology data is available, that should be used in lieu of this system. However, if other regulatory agencies have already set exposure limits, those should be used first. Often they are science-based, they have undergone peer review, and they can be used with confidence. Therefore, the first step is really to see if other regulatory agencies have already set a limit, and if so, use that. The second step is to see if tox data is available, and if so, go ahead and set a tolerable intake level based on the tox data. Only if you don t have those other two pieces of information would you use the TTC approach as a default.

5 To consider how well the TCC values serve to protect against toxicity, see the comparison of the TTC values for the Escher paper with Carthew, Munro, etc. in the table above. With Toluene, where the patient would be exposed to.017 mg/m 3 and the default safe level of exposure that we would estimate from the Escher paper with Carthew, Munro, it was not as protective. In this case, there was actually an EPA reference concentration, and that would be used over the much more protective TCC value. It is about a 3, 4 magnitude difference. This kind of exercise is useful, keeping in mind that the compounds that are used may not be representative of the ones that would come out of medical devices. Therefore, adopting the proposed inhalation TTC values is one proposed solution, but it is a work in progress. More work is needed to decide which analyses make the most sense for medical devices. Applications to mixtures In terms of applying this to mixtures, earlier work at the International Life Sciences Institute explored the potential for interactive effects to occur among mixture constituents, and whether or not this was a dose dependent phenomenon. The collective data indicated that interactive effects could appear at low doses, even at NOAELs, which is not necessarily intuitive. Take as an example two compounds, A and B, which taken separately do not produce an effect at a certain dose. Interestingly, a binary mixture of those compounds, at those same NOAELs, might still produce an interactive effect, either additive or synergistic. However, looking well down in the level below the NOAELs in the neighborhood of the TTC values, in µ/per day the data seems to suggest that the potential for interactive effects to occur is very low. Thus, in terms of TTC the potential for synergistic effects to occur is low. A number of investigators have showed this, including Victor Feron. 2 Alan Boobis and colleagues conducted a similar exercise, 3 showing that when synergy does occur, the magnitude of synergy at low doses did not exceed the levels predicted by additive models by more than a factor of 4. These investigations support the idea that TTCs can be used with confidence, with little potential for interactive effects to occur that are going to be synergistic. For practical purposes, the mixture issue is addressed in the ISO standard, Annex B, which provides guidance on adding up the potencies of the individual compounds. If, for example, you have 15 compounds coming out of a device, the first step is to come up with a tolerable intake for each compound, and then determine how much the patient is exposed to the dose. They are then compared in a hazard index. The sum of the dose to TI ratios are taken, and then added up. For every individual compound, if the hazard index is above 1, there might be cause for concern about the potential for adverse effects to occur. If it is below 1, there is no cause for concern. Keeping in mind that TTC is used when there is insufficient toxicity data to come up with tolerable intake (TI), using this method TI is replaced with TTC. 2 Feron et al. Toxicology 105 (1995): pp Boobis et al. Critical analysis of literature on low-dose synergy for use in screening chemical mixtures for risk assessment. Critical Review in Toxicology. (2011): pp

6 Case study Susan Felter of Procter & Gamble conducted a similar project to evaluate how to apply the TTC approach to compounds in a complex mixture looking at water contaminants as a hypothetical example. She looked at the potential water concentration, and made a default assumption about how much people drink in a day to come up with an exposure. She then ran the ToxTree model, came up with Cramer classes, and then found the TTC values that are appropriate for non-cancer effects. Then, she derived a hazard quotient, hazard index, which is just the ratio of the dose and the TTC. In every case, they were below one. Individually, there is no concern and when you add them up the sum is still below 1, so there d be very little concern about interactive effects to occur for the compounds in this complex mixture. ICH is going in a slightly different direction. They are developing an approach that, regardless of how many compounds are in the mixture, if its additive or synergistic assigns defaults that they think are protective for mixtures. For individual compounds, the TTC is going to be 1.5 for carcinogenic compounds that have lifetime exposure, and ICH bumps it up to 5 with multiple impurities. That s a practical way to do it without worrying about all the individual constituents. However, this may not be very satisfying, and it may be under reconsideration this is still a work in progress. From the perspective of the medical device community, there is a way forward using the hazard index approach that has already been written it into the standards. Once ICH further develops and evolves their approach on defaults it might be considered as well. ICH M7 Draft Guidance: TTC values for mixtures

7 Applicability of the TTC approach to metal-containing compounds Robert Kroes 4 is one of the main investigators developing the TTC approach to metals or metal-containing compounds. Kroes developed a flow chart to guide the choice of TTC values based on whether the compound is carcinogenic or not, toxic or not, etc. One of the criteria that will kick you out right away is whether or not the compound is a metal or contains metal. Also, there are some other bad actors, if it s a dioxin, for example, TTC is not an appropriate approach because those are very potent carcinogens, but they might lie on that tail, that fifth percentile, that TTC is not intended for. Metal compounds are going to be a problem when you apply the TTC approach. Brown served as the co-chair of an ILSI Research Foundation committee that is developing a draft framework for conducting risk assessments for inorganic, organo-metallic, and metallic compounds. The committee looked at dose response data for a large number of compounds, and found that for many of the metallic compounds, the inorganic compounds, regulatory agencies have already set exposure limits for a lot of these compounds. EPA and other regulatory agencies already have exposure limits for mercury, lead, tin, etc. Although a lot of those are based on oral exposure, with appropriate route-to-route extrapolation techniques these can be used with medical devices. The challenge is what to do about organo-metallics. Unlike inorganic metallic compounds, organo-metallics need an alternate approach, because exposure limits are established for only a few compounds. Organometallics are an issue from a medical device perspective because of certain disinfectants and many color additives contain metals. Can the TTC approach be applied to compounds that have metals? In an ideal world, the same approach used for oral compounds and inhalation compounds could be applied, taking a distribution of potency values and assigning a cutoff and a safety number. However, given the lack of data on organo-metallic compounds, does the existing TTC value of 90µg/per day work for these types of compounds? The table shows a snapshot of organo-metallic compounds with NOAEL values identified from the EPA toxicity reference database. Note that if a model is used that tries to identify the Cramer class, it will automatically kick an organo-metallic compound into Cramer class III, because it is a metal. Interestingly, a 4 R. Kroes et al. The threshold of toxicological concern concept in risk assessment. Toxicological Science 86(2) 2005: pp

8 number of the QSAR models will not work with metal-containing moieties, so that is something to keep in mind in comparing the predictive results of the different SAR packages. After identifying the NOAELs in these compounds, Brown derived a provisional reference dose by using a modifying factor of 100. He then compared them against the TTC values (using the TTC value for non-cancer effects for the most potent class of compounds, Cramer Class III, converted to mg/kg/day using a default 70 kg body weight. For these compounds, the 90 µg/day default held up pretty well; the TTC values were adequately protective. The only compound that fails is tributyltin chloride. Brown also looked at exposure limits that various regulatory agencies have set, to check how they compare to the TTC values. Looking at reference doses from EPA, from the high risk database, and NOAEL risk levels from the Agency for Toxic Substances and Disease Registry (ATSDR) in the US, for some of these TTC value is not adequately protective at 90 µg/day. For example, lead in particular has a very low RfD, and the TTC value is orders of magnitude higher than that. This may not be a realistic example of an organometallic compound in a medical device, but if there were a patient exposure to tetraethyl lead, the TTC value would not be protective. In a similar a case study for organo-tins, the TTC value is appropriate if you use the lower number, the 1.5 µg/day. Looking at the broader database of all of these compounds, when you apply the TTC at 1.5 µg/day, it works for all these organo-metallic compounds except lead. If this approach is to be applied to organo-metallics, it might be necessary to take lead compounds off the books. Next, Brown fit these approaches together in a proposed decision framework, asking a series of questions: 1. Is the compound an inorganic metallic compound? If not, use the appropriate risk assessment approach for organic compounds. 2. Has another regulatory agency already derived an exposure limit? If so, go ahead and use that. 3. Does tox data exist for the compound? If yes, go ahead and derive a compound-specific exposure limit. 4. Are we able to derive a compound-specific exposure limit? If yes, then go ahead and use that data from good, reliable, repeat dose studies. Only when the answer to all of those questions is no, when the data does not exist already to derive a value and a regulatory agency has not already established an exposure limit, does the TTC approach kick in. This approach provides a way forward for organo-metallics, though it will undergo some more validation exercises before it is incorporated more broadly. What kind of endpoints is the TTC approach is applicable for; is it applicable for local effects? In a biological evaluation, a wide range of endpoints must be considered, depending on the nature of exposure to the device and the duration. Essentially all devices undergo cytotoxity testing, irritation and sensitivity testing. Where can the TTC approach come in as an alternative to animal tests? Given that the existing data came from systemic toxicity and carcinogenicity studies, it makes the most sense to use the TTC as an alternative to chronic toxicity and carcinogenicity studies of medical devices. As a practical matter, that is where it is the most likely to be used. It is primarily considered only as an alternate approach to resourceintensive, expensive tests. There are still a number of other endpoints that need to be considered in a biological evaluation of a device, including genotoxicity, implantation, and hemocompatibility. With reproductive and developmental effects, some papers in the literature show that the existing TTC values are protective. That s important to keep in mind.

9 Although some have considered that TTC values may be protective enough to use in lieu of genotoxicity tests; that seems premature. There is the potential for this approach to be used as an alternative for in vivo micronucleus testing as part of the genotoxicity testing, but that is not policy anywhere yet. This is especially true for local effects. Depending on your device, that might be your most clinically relevant effect, and these TTC values are not intended to be protective for local effects. They cannot be used as a surrogate for local effects testing whether it is irritation, or implantation. That s important to keep in mind. The TTC approach cannot serve as a fix-all alternative for all testing. It is a powerful tool, and the numbers are science-based, so there is a considerable comfort level associated with them, but it cannot cover testing of all different endpoints. It cannot be an alternative to cytotoxicity testing, which is pretty sensitive, and might have effects at doses lower than the TTC values. TTC may be a very valuable tool, but its not a panacea for all of these issues, and it is not going to be used as an alternative for evaluating local effects, and many other biocompatibility endpoints that we need to consider. Summary There is a paradigm shift underway. Rather than automatically testing the biocompatibility of medical devices in animal models, there are some alternatives to consider. One approach that is increasingly used is the chemical characterization risk assessment approach. But the practical problem in implementing it is that we do not have toxicity data for a lot of the compounds that are released from device materials. That is when the TTC approach may be considered. The way we think about TTC is evolving. It is heading in the direction of not just being used as a way to decide whether more testing is needed, but to actually being used as default TI values. There are some unique considerations to keep in mind when applying the TTC approach to compounds that are released from device materials: the duration of exposure, the route of exposure, mixtures, metal containing compounds, and applicable endpoints. A lot of progress is being made, and real-world practical examples of how to apply the TTC approach, such as the example Richard Hutchinson shared in his presentation, will be very helpful. Such practical examples are useful not only in validating the approach, and encouraging people to think about this approach, but also in addressing science concerns such as the inhalation values. Discussion Need for a pragmatic approach The participants agreed that there is a need for a pragmatic approach, that suggests which purposes we can use the TTC approach for e.g. subacute, sub-chronic toxicity, and chronic toxicity. We should ask also if we want to include genotoxic endpoints. We also need to work on potential in vitro tests for subacute, sub-chronic toxicity. This group hopes that this is the way forward, and good progress is being made with this draft standard on how to actually implement this approach for compounds released from medical devices. This approach has been discussed at various TC 194 meetings, with international buy-in, and now we want to get this down on paper. Also, in terms of applicability to certain endpoints, you really need to be a trained toxicologist and use professional judgment. Some submissions that use this thought process appropriately, others that do not. The purpose of guidance document right now is for people who do know how to use this to break down where it is appropriate, and who understand the limitations. Discussion is crucial at this point, and so is getting the guidance document out is crucial, or else some people will continue with this mixture and inappropriate use and the risk is that it could get a bad reputation. Hypothetical example with lifetime rodent oral dosing study A hypothetical example taking TTC data and a lifetime rodent oral dosing study imagined a compound with a known absorption rate of 33 percent. In that case, should a risk assessment use the TTC 1.5 µg number, or cut that to 0.5 µg for instance, given the absorption rate as an example of the application of professional judgment recommended in the forthcoming guidance document? Once you get to the point where you have a

10 specific CAS number, searched the literature and obtained an absorption rate, you re probably going to have a lot more specific information than a TTC value. We first look for other reference values and approved limits in the guides first, and then look in the literature for established limits, and then lastly consider the TTC value, so in that hypothetical it seems unlikely that you d have ADME data but not robust tox data, so you probably would not even be worried about TTC. If you ve got the data, definitely use bioavailability to adjust it for your route of exposure. Local v. systemic effects These documents were based on TTC oral absorption from food additives, primarily oral exposure. Various suggestions have been proposed for deriving values for inhalation. We re just in the process of sorting through this. The ICH draft guidance believes it is applicable and sufficiently protective. There are TTC values that are derived from inhalation, not oral, data. That data applies to systemic effects, not local effects, except for the Carthew paper, which considered local respiratory effects as well. The Escher and Carthew paper provide more details on how they came up with their numbers. For now, the TC-194 working group document will go with these values.